Smithsonian Institution Scholarly Press smithsonian contributions to zoology • number 635

Smithsonian Institution Scholarly Press

Variation,A Chronology Systematics, of and MiddleRelationships Missouri of Plainsthe Leptodactylus bolivianus ComplexVillage (Amphibia: Sites Anura: Leptodactylidae) By Craig M. Johnson

W. Ronaldwith Heyer contributions and Rafael by O. de Sá Stanley A. Ahler, Herbert Haas, and Georges Bonani Series Publications of the Smithsonian Institution

Emphasis upon publication as a means of “diffusing knowledge” was expressed by the first Secretary of the Smithsonian. In his formal plan for the Institution, Joseph Henry outlined a program that included the following statement: “It is proposed to publish a series of reports, giving an account of the new discoveries in science, and of the changes made from year to year in all branches of knowledge.” This theme of basic research has been adhered to through the years by thousands of titles issued in series publications under the Smithsonian imprint, com- mencing with Smithsonian Contributions to Knowledge in 1848 and continuing with the following active series:

Smithsonian Contributions to Anthropology Smithsonian Contributions to Botany Smithsonian Contributions to History and Technology Smithsonian Contributions to the Marine Sciences Smithsonian Contributions to Museum Conservation Smithsonian Contributions to Paleobiology Smithsonian Contributions to Zoology

In these series, the Institution publishes small papers and full-scale monographs that report on the research and collections of its ­various museums and bureaus. The Smithsonian Contributions Series are distributed via mailing lists to libraries, universities, and similar institu- tions throughout the world.

Manuscripts submitted for series publication are received by the ­Smithsonian Institution Scholarly Press from authors with direct affilia- tion with the various Smithsonian museums or bureaus and are subject to peer review and review for compliance with manuscript preparation guidelines. General requirements for manuscript preparation are on the inside back cover of printed volumes. For detailed submissions require- ments and to review the “Manuscript Preparation and Style Guide for Authors,” visit the Submissions page at www.scholarlypress.si.edu. smithsonian contributions to zoology • number 635

Variation, Systematics, and Relationships of the Leptodactylus bolivianus Complex (Amphibia: Anura: Leptodactylidae)

W. Ronald Heyer and Rafael O. de Sá

washington d.c. 2011 ABSTRACT Heyer, W. Ronald, and Rafael O. de Sá. Variation, Systematics, and Relationships of the Lepto- dactylus bolivianus Complex (Amphibia: Anura: Leptodactylidae). Smithsonian Contributions to Zoology, number 635, viii + 58 pages, 21 figures, 20 tables, 2011.—The Leptodactylus bolivianus complex has been considered to consist of one or two species, L. bolivianus alone or L. bolivianus and L. insularum. Detailed morphological analyses were undertaken to evaluate variation in the complex, which ranges from Costa Rica through Panama, across northern South America in the river valleys draining to the Caribbean, and throughout much of the Amazon basin with southern limits in Bolivia. Members of the complex also occur on several islands off Nicaragua, Panama, and Colombia. Analyses of morphological and advertisement call data indicate that there are either two or three species comprising the complex. Analysis of molecular data strongly supports recognition of three species, one of which is described as a new species, Leptodactylus guianensis. The three species comprising the L. bolivianus clade are most closely related to the L. ocellatus clade within the genus Leptodactylus.

Cover photos, from left to right: Leptodactylus insularum (Figure 17, courtesy of R. W. Mc- Diarmid), L. guianensis (Figure 15, courtesy of M. S. Hoogmoed), and L. bolivianus (Figure 12, courtesy of R. B. Cocroft).

Published by SMITHSONIAN INSTITUTION SCHOLARLY PRESS P.O. Box 37012, MRC 957 Washington, D.C. 20013-7012 www.scholarlypress.si.edu

Text and images in this publication may be protected by copyright and other restrictions or owned by indi- viduals and entities other than, and in addition to, the Smithsonian Institution. Fair use of copyrighted mate- rial includes the use of protected materials for personal, educational, or noncommercial purposes. Users must cite author and source of content, must not alter or modify content, and must comply with all other terms or restrictions that may be applicable. Users are responsible for securing permission from a rights holder for any other use.

Library of Congress Cataloging-in-Publication Data Heyer, W. Ronald. Variation, systematics, and relationships of the Leptodactylus bolivianus complex (Amphibia: Anura: Leptodactylidae) / W. Ronald Heyer and Rafael O. de Sá. p. cm. — (Smithsonian contributions to zoology ; number 635) Includes bibliographical references. 1. Leptodactylus bolivianus—Classification. 2. Leptodactylus bolivianus—Variation. I. De Sá, Rafael O. II. Title. QL668.E257H49 2011 597.8’75—dc22 2011004864

ISSN: 0081-0282 (print); 1943-6696 (online)

The paper used in this publication meets the minimum requirements of the American National Standard for Permanence of Paper for Printed Library Materials Z39.48–1992. Contents

LIST OF FIGURES  v

LIST OF TABLES  vii

INTRODUCTION  1

METHODS AND MATERIALS  1 Juvenile and Adult Morphological Data 2 Larval Morphology 3 Advertisement Calls 4 Molecular Sequence Data 5

Measurement DATA  5

INTRASAMPLE VARIATION  6 Dorsal Pattern 8 Subocular Spot 9 Lip Stripe 9 Belly Pattern 9 Thigh Pattern 9 Shank Pattern 9 Dorsolateral Folds and Lateral Folds 9 Male Secondary Sexual Characters 9 Measurement Data 9

INTERSAMPLE VARIATION  12 Patterns and Folds 12 Dorsal Pattern 12 Subocular Spot 14 Lip Stripe 17 Belly Pattern 19 Thigh Pattern 19 Shank Pattern 20 Dorsolateral Folds 21 Lateral Folds 21 iv • smithsonian contributions to zoology

Male Secondary Sexual Characters 21 Measurement Data 22

GEOGRAPHIC VARIATION AMONG THE SAMPLES  23 Adult Form Morphological Characters 23 Larval Morphology 24 Advertisement Calls 25 Costa Rica Recording 25 Panama Recording 25 Peru Recording 25

MOLECULAR RELATIONSHIPS  26

INTERSPECIFIC VARIATION  27 Measurement Data 28 Morphological Data 29 Leptodactylus bolivianus and L. insularum Comparisons 30 Leptodactylus bolivianus: New Species Comparisons 30 Leptodactylus insularum: New Species Comparisons 30

SPECIES ACCOUNTS 30 Leptodactylus bolivianus Boulenger, 1898 31 Leptodactylus guianensis, new species 35 Leptodactylus insularum Barbour, 1906 38

ACKNOWLEDGMENTS 45 APPENDIX 1: OBSERVER ERROR ASSESSMENT FOR MORPHOLOGICAL DATA 47

APPENDIX 2: molecular sample LOCALITY DATA 55

REFERENCES 57 Figures

1. Locations of specimens used for intrasample analyses 7 2. Locations of 20 regional samples 17 3. Variation in posterior thigh patterns among 20 regional samples 20 4. Female discriminant function analysis results of 20 geographic samples 23 5. Levels of differentiation between geographically adjacent samples 23 6. Advertisement call of Leptodactylus from Costa Rica 26 7. Advertisement call of Leptodactylus from Panama 27 8. Advertisement call of Leptodactylus from Peru 28 9. Maximum likelihood cladogram of the Leptodactylus ocellatus species group 29 10. Plot of first two canonical vectors for male measurement data 29 11. Plot of first two canonical vectors for female measurement data 30 12. Leptodactylus bolivianus from Tambopata, Peru 32 13. Male thumb spines 32 14. Distributions of Leptodactylus bolivianus, L. guianensis, and L. insularum 34 15. Leptodactylus guianensis from northern Pará, Brazil 35 16. Lateral view of head, dorsal and ventral views of Leptodactylus guianensis, new species 36 17. Leptodactylus insularum from San Andrés, Colombia 38 18. Dorsal and ventral views of the lectotype of Leptodactylus insularum 39 19. Lateral view of Leptodactylus insularum tadpole 40 20. Oral disk of Leptodactylus insularum tadpole 40 Appendix 1 figure A1.1. Example of design pattern 50

Tables

1. Statistics to determine observer error 6 2. Dorsal pattern state occurrences within Lower Río Chucunaque sample 8 3. Dorsal pattern state occurrences by age and sex within Lower Río Chucunaque sample  8 4. Chi-square occurrence values for character states of the dorsal pattern within Lower Río Chucunaque sample 8 5. Age and sex intrasample analyses results for dorsal pattern 8 6. Dorsal pattern, suborbital spot, lip pattern, and belly pattern state occurrences 10 7. Age and sex intrasample analyses results for subocular spot 11 8. Age and sex intrasample analyses results for lip stripe pattern 11 9. Age and sex intrasample analyses results for belly pattern 12 10. Age and sex intrasample analyses results for thigh pattern 12 11. Thigh pattern, shank pattern, dorsolateral fold, and lateral fold occurrences 13 12. Age and sex intrasample analyses results for dorsolateral folds 14 13. Age and sex intrasample analyses results for lateral folds 14 14. Male secondary sexual character state occurrences 15 15. Intrasample variation in measurement data 16 16. Intersample comparisons for dorsal pattern data 18 17. Number of males exhibiting secondary sexual character states 22 18. Quantitative larval data 24 19. Advertisement call data 25 20. Morphological character states for species pairs of the Leptodactylus bolivianus complex 31

Variation, Systematics, and Relationships of the Leptodactylus bolivianus Complex (Amphibia: Anura: Leptodactylidae)

INTRODUCTION

A cluster of morphologically similar frogs of the genus Leptodactylus hav- ing a pair of distinct dorsolateral folds on the dorsum and well-­developed lat- eral fringes on the toes has never been systematically evaluated by examining materials from throughout its geographic range. The species involved are herein referred to as members of the Leptodactylus bolivianus complex. There have been three names proposed for members of this complex: Leptodactylus bolivi- anus Boulenger, 1898; Leptodactylus insularum Barbour, 1906; and Leptodac- tylus romani Melin, 1941. The collective range for the L. bolivianus complex is from Costa Rica southward through Panama, extending across northern South America (east of the Andes) in the river valleys draining to the Caribbean, and throughout much of the Amazon basin with southernmost limits in the Bolivian departments of La Paz, Cochabamba, and Santa Cruz. We analyze variation in this complex of frogs throughout its geographic range to understand inherent patterns of differentiation and to interpret those patterns in terms of species-­level recognition, distributions, and relationships.

METHODS AND MATERIALS W. Ronald Heyer, Division of Amphibians and Reptiles, National Museum of Natural His- We follow Fabrezi and Alberch (1996) in numbering the fingers II–V (con- tory, Smithsonian Institution, MRC 162, P.O. sidering finger I as lost in frogs). Herein, we use the name Leptodactylus latrans Box 37012, Washington, D.C. 20013-­7012, USA. Rafael O. de Sá, Department of Biology, following the recent elucidation of the nomenclature of the widely spread taxon University of Richmond, Richmond, Virginia previously referred to as L. ocellatus (Lavilla et al., 2010). 23173. Correspondence: [email protected]. Manu- The primary materials examined for this study are juvenile and adult script received 28 June 2010; accepted 23 Sep- museum specimens; only this data set provides extensive locality records for tember 2010. the members of this complex. The juvenile and adult morphological data are 2 • smithsonian contributions to zoology complemented by available data on larval morphology, ULABG Universidad de Los Andes, Laboratorio advertisement calls, and molecular sequences. de Biogeografía, Mérida, Venezuela The names of the institutions and collections from The following are field tag abbreviations, not yet in- which specimen data were obtained are abbreviated here corporated into collections. (after Sabaj Pérez, 2010) for subsequent mention in the text. AMU Abraham Mijares ELM Enrique La Marca AMNH American Museum of Natural History JC Jay Cole CM Carnegie Museum of Natural History RdS Rafael de Sá FMNH Field Museum of Natural History UC Ulisses Caramaschi ICNMHN Instituto de Ciencias Naturales, Museo USNM-FS­ National Museum of Natural History, de Historia Natural, Universidad Nacio- Smithsonian Institution, Washington, D.C. nal de Colombia IND Instituto Nacional de Recursos Naturales Juvenile and Adult Morphological Data Renovables y del Ambiente, Bogota INPA Instituto Nacional de Pesquisas da Ama- The data involved were taken and analyzed by the first zonia, Brazil author (WRH) for consistency (see Hayek et al., 2001). KU University of Kansas Natural History Because of the moderately large size of adult specimens Museum of this complex and new restrictions on transportation of LSUMZ Louisiana Museum of Natural History alcohol-­stored specimens, it was necessary to take data on MCZ Museum of Comparative Zoology, Har- the specimens in the collections involved rather than bor- vard University rowing the specimens. Thus only the USNM specimens, MNCN Museo Nacional de Ciencias Naturales, plus a few other specimens on loan to WRH, were avail- Madrid able for reexamination during this study. MNRJ Museu Nacional, Universidade Federal Specimen data from the following collections were do Rio de Janeiro taken in situ: AMNH, FMNH, ICNMHN, IND, KU, MPEG Museu Paraense “Emilio Goeldi,” Belém, MCZ, MNRJ, Museo Universidad Javariana–Bogota, Brazil MZUSP, UMMZ, and USNM. MPUJ Museo de Pontificia Universidade de Ja- Geographic samples are analyzed to evaluate the na- veriana, Bogotá ture of morphological variation. The samples are indicated MZUSP Museo de Zoologia da Universidade de by italics and leading capital letters. São Paulo, São Paulo In 2005 an evaluation of the assembled locality data NRM Naturhistoriska Riksmuseet, Department was undertaken to determine if there were any obvious of Vertebrate Zoology, Stockholm geographic gaps for which specimens should be sought TCWC Texas Cooperative Wildlife Collection, in other collections. There was only one obvious gap— Texas A&M University, College Station, no data were available from the Brazilian State of Pará, Texas even though members of the complex occur in extensive TNHC Texas Natural History Collections, Texas portions of Amazonia. The MZUSP and USNM collec- Natural Science Center, Texas Memorial tions contain rather extensive amphibian collections from Museum, University of Texas at Austin Pará, suggesting that members of the complex really do UMMZ University of Michigan Museum of Zool- not occur in Pará. The institution housing the most her- ogy, Ann Arbor, Michigan petological specimens from Pará is the MPEG. Dr. Mari- USNM National Museum of Natural History, nus Hoogmoed reported that there were no specimens of Smithsonian Institution, Washington, D.C. the L. bolivianus complex catalogued in the MPEG col- UTA University of Texas at Arlington, Depart- lection (M. Hoogmoed, personal communication [e-mail]­ ment of Biology, Texas to WRH, 30 March 2005). At the VI World Congress of The following collections are not included in Sabaj Herpetology in July 2008, Dr. Hoogmoed (pers. comm.) Pérez (2010). informed WRH that he had reexamined frogs from the state of Pará, near the state of Amapá and found them AL Adolfo Lutz collection housed at MNRJ to be members of the Leptodactylus bolivianus complex. number 635 • 3

Subsequently (e-mail­ message of 20 December 2008 to to express the completeness of the pair of dorsolateral WRH) Dr. Hoogmoed wrote, “During the last fieldwork folds extending from the supratympanic fold to the groin again 4 specimens of Leptodactylus bolivianus [new spe- along the junction of the dorsal and flank regions and the cies as recognized in this paper]. Thus the species is well pair of lateral folds extending from the supratympanic distributed in Para north of the Amazon.” fold to the groin in the midflank region. Specimens from the collections indicated above should (4) Male secondary characteristics. Combinations of be adequate to evaluate patterns of morphological varia- numbers, letters, pluses, minuses, and parentheses were tion throughout the distribution of the species complex. used to code variation in male arm hypertrophy, tympanic After examination of materials from the AMNH, tubercles, jaw tubercles, chest tubercles, and thumb spines. ICNMHN, IND, MCZ, MNRJ, MZUSP, and USNM, it (5) Measurements. Measurement data were taken seemed that data taken to that point on surface textures with calipers either with the unaided eye or under a dis- of the dorsal shank, outer tibia, and sole of foot did not secting microscope for snout–vent length (SVL), head demonstrate meaningful interpopulational variation. Be- length, head width, shank length, and foot length follow- fore visiting further collections, an analysis was made (see ing Heyer (2005). Measurements on MZUSP specimens Appendix 1) that confirmed the uninformative value of were recorded to the nearest 0.5 mm; all others were re- those characters, and no further data were taken for these corded to the nearest 0.1 mm. variables. Observer error increases when materials are not The variables recorded for the entire data set for anal- available for reexamination. Therefore this problem was ysis of variation organized by category are as follows: specifically addressed before the data were analyzed for (1) Sex. Adult males are defined as having vocal slits. geographic variation. The methods used to determine ob- Juvenile males lack vocal slits. Gonads of individuals near server error are described in Appendix 1. adult size were examined to determine whether they were juvenile males or females. Adult females are defined as hav- Larval Morphology ing at least curly oviducts. Juvenile females have straight oviducts. Most small specimens were not dissected to ex- Terminology and methods follow McDiarmid and amine for gonads and are classified as juveniles. Altig (1999) including use of Gosner developmental (2) Belly, dorsal, lip, posterior thigh, and dorsal shank stages. In addition to the examined USNM larval collec- patterns. Outline drawings were filled in for each of these tion for samples of this complex, AMNH, CM, FMNH, characters, and these drawings served as pattern stan- ICNMHN, KU, MCZ, and UMMZ were contacted for dards; additional standards were created as deemed ap- larvae identified as either L. bolivianus or L. insularum in propriate. For data-­recording purposes, each pattern was their collections. given a unique letter or combination of a letter and num- There are very few larval samples from the above col- ber. Specimens that had patterns intermediate between lections. Only KU and USNM had samples identified as two standards where creation of a new standard was con- belonging to the L. bolivianus-­insularum complex. The sidered inappropriate were recorded as having both pat- KU samples were borrowed and examined by WRH. terns separated by a slash. These slashed pattern notations The KU larval specimens from two localities have are treated as separate states. Belly patterns were based been described in the literature (Duellman, 1997, 2005). on standards for the Leptodactylus pentadactylus species Upon examination of these specimens, WRH determined group (Heyer, 2005) as a starting point to which standards that the published descriptions were based on misidenti- were added as needed. The dorsal, lip stripe, posterior fied specimens. Duellman (1997) described KU 167784 thigh, and dorsal shank patterns used were initially cre- and 167785 from km 104 on the road from El Dorado to ated on the basis of small samples of specimens of both Santa Elena de Uairén, Bolívar, Venezuela, as L. bolivianus. the L. bolivianus and L. latrans clusters with additional No juvenile or adult KU specimens of the L. bolivianus-­ standards added in this study. The first two collections ex- insularum complex were collected at this locality. The tad- amined were MZUSP and USNM during which the major- poles are in fair condition, very soft, with all teeth missing. ity of new standards were created. Subsequently, any new Duellman (1997) indicated that seven individuals at Gos- standard created was dated to know when in the study it ner stage 25 had body lengths of 15.5–18.4 mm. Data was available for comparison. were taken on five larvae from each sample that visually (3) Dorsolateral and lateral folds. Combinations of appeared to cover the size ranges for each sample. Tad- letters, numbers, and parentheses were used in data coding pole stages ranged from 26 to 35 with body sizes ranging 4 • smithsonian contributions to zoology from 14.0 to 18.1 mm. Duellman (1997) did not report a There are three commercially available CDs contain- tooth row formula for the larvae. Two tadpoles were ex- ing purported recordings of specimens of the L. bolivianus-­ amined using crystal violet stain to differentiate the tooth insularum complex. row ridges that bear the teeth. The anterior ridges were Márquez et al. (2002) contains two brief record- unclear in one individual, but the other individual has a ings of advertisement calls identified as L. bolivianus. 3(3)/3(1) formula. Both tadpoles have the same posterior The recordings were obtained near a pond where several tooth row condition, with a short P3 row, which is un- males were calling, and no particular voucher was col- usual for Leptodactylus larvae but does occur in a few lected (Ignacio De la Riva, pers. comm. [e-mail],­ 6 July members of the L. pentadactylus species cluster, such as in 2006). These calls from Bolivia are very different from some individuals of L. knudseni (Heyer, 2005). The vent those recorded from Tambopata, Peru, for which there is appears to be median, although the condition of preser- a museum voucher specimen corresponding to L. bolivi- vation makes the evaluation difficult. Given the 3(3)/3(1) anus (USNM 343347). Either the Bolivian recording spe- tooth row formula, the larvae most likely do not belong to cies identification is in error or the Bolivian and Peruvian the genus Leptodactylus. Duellman (2005) described and frogs associated with the calls represent distinct species. illustrated larvae of L. bolivianus from Cusco Amazonico, The Bolivian recordings have considerable sound in the Madre de Dios, Peru. The specimen illustrated in Duell­ same broadcast channel as the call of interest and cannot man (2005:177, fig. 13.10A) was placed in a separate be analyzed in much detail. On the basis of the available container and identified as the illustrated specimen. There information, these Bolivian CD recordings are considered is one discrepancy between the specimen, Duellman’s de- to not belong to a member of the L. bolivianus-­insularum scription, and the illustration. Duellman (2005:279) de- complex. scribed a specimen at stage 25. The illustration shows a The Cusco Amazónico recording KU cassette 104/3 limb bud with early differentiated digits, perhaps at stage published and figured by Duellman (2005:279) for L. bo- 35 or 36, but not of normal shape for digits at those stages livianus was requested for analysis. KU cassette 104/3 of development. Duellman (2005) reported the tooth row was mistakenly discarded before it was incorporated formula as 2/3(1). Reexamination of the illustrated speci- into the KU tape archives (W. E. Duellman, pers. comm., men indicates a formula of 1/3(1), with a very short P3 e-­mail message of 27 June 2006). The description pro- row. Examination of several specimens from KU 205810 vided (Duellman, 2005:279) could represent either L. bo- indicates the variable formula 1/2–3(1), with P3 always livianus or more probably (on the basis of the statement, very short when present. A single tadpole was exam- “The call consists of a continuous bubbling sound”), ined using crystal violet stain. A broadly separated pair L. leptodactyloides. of ridges with two lateral extensions is present where an Two Costa Rican recordings were borrowed for A2 row would be expected. As indicated, none of these analysis: KU 801 and KU 802. The KU 801 recording has ridges bear teeth in the several specimens examined from minor tape speed problems. The KU 802 recording has lot KU 205810. The tadpoles identified as L. bolivianus major tape speed problems and was not analyzed. (Duellman, 2005) do not belong to the genus Leptodacty- Three recordings were analyzed with Raven 1.2 soft- lus; most likely, they are unidentified larvae of leiuperid, ware (Charif et al., 2004); terminology follows Heyer et leptodactylid, or dendrobatid frogs. al. (1990). Data for beginning, highest, and ending fre- quencies were taken from audiospectrogram displays. Advertisement Calls Dominant frequencies were obtained from spectrum anal- yses of entire calls. Call durations were taken from wave- The same institutions that were contacted for larval form displays. All calls were filtered around the broadcast holdings were contacted for holdings of any recordings of frequencies for analysis. Default settings were used for au- advertisement calls. Two recordings were available at the diospectrograms (window: Type Hann, size 512 samples, Texas Natural History Collection. Both recordings were 3 dB filter bandwidth 124 Hz; time/grid: overlap 50%, from the canal zone region of Panama, one of poor qual- hop size 256 samples; frequency grid: DFT size 512 sam- ity. The better recording was not requested, as there was ples, grid spacing 86.1 Hz; averaging: 1 spectra) and spec- another recording available for the same area. In addition, trums (window: size 256 samples, 3 dB filter bandwidth César Barrio and Enrique La Marca reported that they did 248 Hz; time grid: overlap 50%, hop size 128 samples; not know of any recordings of the L. bolivianus-­insularum frequency grid: DFT size 256 samples, grid spacing 172 complex from Venezuela. Hz; averaging: 1 spectra). number 635 • 5

Molecular Sequence Data parsimonious trees resulted from the MP searches. Maxi- mum likelihood analyses used the general time-reversible­ Total genomic DNA was extracted using DNAeasy model with empirical base frequencies, gamma distribu- kit (USB) from liver or thigh muscle that was preserved tion of across-­site rate variation, and an estimated propor- in ethanol 95%. A segment of about 850 base pairs (bp) tion of invariable sites (GTR + Γ + I; Swofford et al., 1996) from the 16S ribosomal RNA (rRNA) gene was poly- recommended by ModelTest 3.04 (Posada and Crandall, merase chain reaction (PCR) amplified (Palumbi, 1996) 1998). Clade support of inferred trees was assessed by using an MJ Research PTC-­200 thermocycler. Double-­ nonparametric bootstrapping (Felsenstein, 1985) on the stranded PCR amplifications were performed using Green basis of 100 pseudoreplicates. Taq Master Mix (Promega). Seventeen available samples from the L. bolivianus-­ Amplification primers used were 16S Ar insularum complex (additional data provided in Appendix 5-­CGCCTGTTTACCAAAAACAT-­3 and 16S Br 2) in addition to a sample of L. latrans and of L. chaquen- 5-­CTCCGGTCTGAACTCAGATCACGTAG-­3 under sis were considered as the ingroup taxa (traditionally the following thermal conditions: initial denaturation at all included in the L. latrans species group). To include 94°C for 2 min followed by 34 cycles of 94°C for 1 min, variation found in the genus Leptodactylus (sensu stricto), 58°C for 1 min, and 72°C for 1 min 30 s. Amplified seg- the following taxa were used as outgroups: L. podicipi- ments were purified using Exo-­Sap (USB) by heating nus, L. wagneri (melanonotus species group), L. mystaci- samples at 80°C for 15 min. Purified products were cycle nus, L. gracilis (fuscus species group), L. paraensis, and sequenced with the dideoxy chain termination method L. knudseni (pentadactylus species group). using the Sequi-Therm­ Excel II DNA sequencing kit (Epi- centre Technologies). Infrared-­labeled sequencing primers (same as amplification primers) were used in sequenc- Measurement Data ing reactions (LiCor Biotechnology) under the following thermal conditions: initial denaturation at 95°C for 2 min The protocol recommended by Hayek et al. (2001) of 30 s, followed by 30 cycles of 95°C for 30 s, 58°C for measuring a single individual male and female 20 times 30 s, and 70°C for 30 s. Sequencing products were run in was followed. For this study, two sets of specimens were 6% acrylamide, 44 cm in length, gels using a Licor DNA remeasured. The first set represents well-­preserved speci- 4300 automatic sequencer. Sequencing reactions were mens USNM 200352 (female) and USNM 200353 (male). single stranded; double-­stranded PCR fragments were Twenty-­one data sheets were prepared to record data (to sequenced in both directions. Sequences included in the allow for at least one unambiguous error to be discarded analyses represent a consensus of both DNA strands; Gen- and still have 20 data observations). Only a single sheet Bank accession numbers for the sequence data are: HQ was filled out on any given day. These data were taken 232831–232846, HQ259119, HQ259120, AY947857, on the same days other data were being taken on USNM AY911285, AY669856, AY943235, EF632048, EF632053, specimens. The sheets were filled in such that male data and EF632055 were recorded first on alternate sessions and the sheets Complementary sequence strands for each sample were filled out alternately at either the beginning of the were first aligned and, using the chromatographs cre- data taking session or the end of the session. Data were ated by BaseImagr software (LicorBiotechnology), were taken between 19 August and 28 October 2004. The sec- inspected visually for mismatches of aligned positions to ond set represents stiff specimens (presumably yielding confirm or manually correct the automatic reading. Con- larger measurement errors): USNM 145708 (male) and sensus sequences were aligned with ClustalX (Thompson USNM 145711 (female). Twenty-­one data sheets were et al., 1997) using the multiple-­alignment option. Phyloge- prepared to record data. A single sheet was filled out on netic analyses of sequence data were run using maximum any given day. Male data were recorded first on alternate parsimony (MP) and maximum likelihood (ML) criteria as recording sessions. Data were taken between 21 Septem- implemented in PAUP* 4b6 (Swofford, 2002). Character ber and 21 October 2005. states were treated as unordered, and gaps were consid- Surprisingly, the well-preserved­ specimens were not ered as a fifth character (MP analysis); heuristic searches notably measured more accurately than the stiff specimens were performed with 100 random additions of sequences (Table 1). For example, half of the well-preserved­ specimen and tree bisection-reconnection­ (TBR) branch swapping. variables had the highest coefficient of variation values. Strict consensus trees were calculated when several equally As it makes little or no difference which male and female 6 • smithsonian contributions to zoology

TABLE 1. Summary statistics for specimens remeasured 21 times to determine observer error. Definitions: Min, minimum measurement; Max, maximum measurement; CV, coefficient of variation; SVL, snout–vent length; HL, head length; HW, head width; EN, eye-nostril distance; TD, tympanum diameter.

Well-preserved specimens Stiffly preserved specimens

♂ USNM 200353 ♀ USNM 200352 ♂ USNM 145708 ♀ USNM 145711 Specimen variable Min Max CV Min Max CV Min Max CV Min Max CV

SVL 87.8 88.9 0.004 87.4 89.3 0.007 87.4 88.5 0.003 75.8 76.8 0.003 HL 30.8 33.3 0.019 31.2 32.2 0.011 30.3 32.3 0.016 26.8 29.3 0.026 HW 31.0 32.1 0.010 29.1 30.9 0.015 30.0 31.1 0.008 26.2 27.5 0.013 EN 8.3 8.7 0.013 8.0 8.6 0.018 8.1 8.4 0.011 7.7 8.0 0.011 TD 6.2 6.6 0.014 6.0 6.3 0.018 5.8 6.3 0.020 5.1 5.6 0.026 Thigh 39.0 39.8 0.005 40.5 42.5 0.013 37.6 38.5 0.007 34.0 35.3 0.009 Shank 39.8 40.8 0.006 43.3 43.8 0.003 42.2 43.3 0.008 39.3 40.2 0.007 Foot 42.2 43.6 0.009 44.4 46.2 0.010 42.5 45.3 0.014 40.4 42.1 0.012

data set is used in subsequent analyses, the male data for Some intrasample data sets therefore comprise data USNM 145708 and female data for USNM 145711 with pooled from nearby localities rather than single localities. the greatest magnitude of differences in coefficient of vari- The following samples are used for intrasample variation ation scores (comparing within sex) are selected for use in analyses (Figure 1): (1) Costa Rica/Panama Border region subsequent analyses. (8 localities, n = 72); (2) Panama Canal region (23 locali- The measurement data have such a low power of res- ties, n = 46); (3) Lower Río Chucunaque, Panama region olution to separate geographic and species samples that (12 localities, n = 58); (4) Ilha do Maracá, Brazil (1 local- inclusion of the multiple measurement data from single in- ity, n = 36); (5) Boca do Acre, Brazil (1 locality, n = 36); dividuals would be superfluous and were not incorporated and (6) Cuzco Amazónico, Peru (1 locality, n = 32). in any discriminant function analyses. The basic approach to analyzing the data in terms of whether the data are combined or analyzed separately by age or sex is to be conservative in combining data. That is, INTRASAMPLE VARIATION if the data show any reasonable evidence for demonstrat- ing variation by age or sex, that is sufficient for intersam- Variation of characters within samples is necessary as ple analyses to be analyzed separately by the appropriate background information to understanding species varia- categories. tion. Also, for the nonsexually dimorphic morphological The dorsal pattern data from the Lower Río Chucu- features, the intrasample data are evaluated to determine naque, Panama are used to explain how the data were ana- whether the characters demonstrate age or sex variation. lyzed. The raw data for this character are shown in Table If the characters vary by either age or sex, the characters 2, where five categories are used to summarize occurrence need to be analyzed separately by those categories. The data for each character state: juvenile, juvenile male, juve- available data for this study do not have large sample sizes nile female, male, and female. As not all nonadult speci- for either intra- ­or intersample analyses. Given that fact, mens were dissected to verify sex, the categories juvenile there is an advantage to combining data by either age or versus juvenile male and juvenile female are not entirely dis- sex in order to increase sample sizes for analysis, when it crete, but overall they represent a size difference with speci- is appropriate to do so. mens categorized as juveniles being in most cases smaller Ideally, the intrasample analyses should be based on than those categorized as juvenile males or juvenile females. specimens from single localities from throughout the geo- The data from Table 2 were combined by either age or sex graphic distribution of the complex. However, there are (Table 3). The observer error analysis for dorsal pattern in- not adequate data from some major geographic areas. dicated that there had to be a greater than 33% difference number 635 • 7

2 3

1 4

5 6

FIGURE 1. Map showing locations of specimens used for intrasample analyses: 1, Costa Rica/Panama Border; 2, Panama Canal; 3, Lower Río Chucunaque, Panama; 4, Ilha do Maracá, Brazil; 5, Boca do Acre, Brazil; and 6, Cusco Amazónico, Peru.

in state distributions to be meaningful. The comparison of with that in mind. In cases where only one character state juvenile with adult data are determined by subtracting the had expected values ≥5, the notation NA is used to indicate smaller percent value from the larger percent value for the that the sample sizes are not appropriate (large enough) for States A, B, and C, as indicted in the formula ((37 − 5) + statistical analysis. For the male–female data comparisons, (13 − 0) + (82 − 63)) = 64%. Thus the dorsal pattern char- the differences among state distributions are ((31 − 23) + acter state distribution for the juvenile/adult categorization (8 − 0) + (69 − 69)) = 16%, below the threshold for ob- exceeds the 33% threshold for observer error. For cases server error and thus are not analyzed further. Such results such as this, chi-­square analyses were performed to deter- are due to two causes, which cannot be accurately sepa- mine whether the differences were statistically significant. rated: (1) the variation reflects actual sample variation and In this case, only two states had expected values ≥5 (Table is not meaningful, or (2) the variation recorded does not 4); accordingly, the chi-square­ analysis was performed only reflect the sample variation but is due to observer error. on the two character states with expected values ≥5. These Decisions on whether it is appropriate to combine analyses are thus conservative and need to be evaluated categories of age or sex are made on evaluation of all of 8 • smithsonian contributions to zoology

the individual sample analyses results. In other words, an TABLE 2. Dorsal pattern state occurrences (by age and sex) with- overall pattern among locality sample results is used to in the Lower Río Chucunaque, Panama sample State A, uniform determine whether categories are appropriately combined between interocular bar (if present) and sacrum; state B, single or not for the intersample geographic variation analyses. chevron between interocular bar and sacrum without additional This approach assumes that there is no geographic varia- spots lateral to dorsal chevron; state C, pattern complex with two tion in whether characters vary by either age or sex (there dorsal chevrons and spotting all over dorsum, chevrons often elongate and fused with each other and interorbital bar; J, juve- are no data at the outset to support or refute this alterna- nile. A dash (–) indicates no specimens demonstrated the pattern. tive). However, at a minimum, it is reasonable to assume that the three Middle American samples would likely demonstrate similar patterns of age or sex category varia- State J J♂ J♀ ♂ ♀ tion as every herpetologist who has studied the Middle A 2 – – 3 4 American members of the L. bolivianus complex in the B 4 1 – – – last 50 years has thought they represent a single species. C 26 4 2 5 7 The following guidelines appear appropriate on the basis of examination of the overall results. If at least five of the samples have variation exceeding observer error and at least one of those samples is statistically significant, the categories being tested should be analyzed separately. If at TABLE 3. Dorsal pattern state occurrences by age and sex, ­Lower Río Chucunaque, Panama sample. Numbers in parenthe- least five of the samples do not exhibit variation exceed- ses are percentages by category (e.g., within juveniles). State A, ing observer error and the one that does is not statisti- uniform between interocular bar (if present) and sacrum; state B, cally significant, the categories can be combined. All other single chevron between interocular bar and sacrum without ad- instances require judgment based on whether chi-square­ ditional spots lateral to dorsal chevron; state C, pattern complex results would likely be significant if sample sizes were in- with two dorsal chevrons and spotting all over dorsum, chevrons creased (to realistic sample sizes). often elongate and fused with each other and interorbital bar. Dorsal Pattern Age Sex The results of analyses by age and sex are shown in State Juvenile Adult ♂ ♀ Table 5. The age results are equivalent in terms of the A 2 (5) 7 (37) 3 (23) 4 (31) B 5 (13) 0 (0) 1 (8) 0 (0) C 32 (82) 12 (63) 9 (69) 9 (69)

TABLE 5. Age and sex intrasample analyses results for dorsal pattern. Definitions: >OE, observer error threshold passed; NS, chi-square results not statistically significant at≥ 0.05 level; NA, TABLE 4. Chi-square occurrence values for character states of data inadequate for chi-square analysis; c/w = compared with. the dorsal pattern, Lower Río Chucunaque, Panama sample. A dash (–) indicates that it would be inappropriate to perform a State A, uniform between interocular bar (if present) and sa- significance test as the observer error threshold was not passed. crum; state B, single chevron between interocular bar and sa- crum without additional spots lateral to dorsal chevron; state C, pattern complex with two dorsal chevrons and spotting all over Juveniles c/w Adults Females c/w Males dorsum, chevrons often elongate and fused with each other and Sample >OE Significance >OE Significance interorbital bar. Lower Río Chucunaque, Yes NS No – Panama Number of juveniles Number of juveniles State observed expected Panama Canal Yes NS No – Costa Rica/Panama Border No – Yes NS A 2 6 Cuzco Amazónico, Peru Yes NS Yes NA B 5 3 Boca do Acre, Brazil Yes NA No – C 32 30 Ilha do Maracá, Brazil No – Yes NA number 635 • 9 guidelines described above. The chi-square­ values for the sample sizes involved is unlikely to result in statistical sig- three samples that exceed the observer error threshold are nificance. The juvenile and adult data are combined. Ap- 0.158, 0.094, and 0.094. It is reasonable to expect that plication of the guidelines for the male and female data these values could reach a 5% significance level if sample analysis results (Table 9) yield combining of these catego- sizes were increased. Therefore dorsal patterns are ana- ries as well for the intrasample summary (Table 6) and lyzed separately by age (juveniles versus adults) in subse- subsequent intersample analyses. quent intersample analyses. Male and female results are also equivalent. The one sample that is appropriate for Thigh Pattern statistical testing has a chi-­square value of 0.371. The two samples that exceed the observer error threshold that are Application of the guidelines to the age and sex analy- not appropriate for statistical analysis could easily have sis results (Table 10) indicates that both age and sex cat- comparable character state distributions with changing egories are best combined for the intrasample summary only 2–4 state conditions. In this case, it is not probable (Table 11) and in subsequent intersample analyses. that increasing sample sizes would lead to statistically sig- nificant results. Male and female dorsal pattern data are Shank Pattern combined for the intrasample variation summary (Table 6) and in subsequent intersample analyses. Application of the guidelines to the age and sex re- sults indicated that both age and sex categories are best Subocular Spot combined for the intrasample summary (Table 11) and in subsequent intersample analyses. None of the six samples Application of the guidelines to the age and sex analy- analyzed met the observer error threshold for either age or sis results (Table 7) indicates that juvenile and adult suboc- sex analysis. ular spot states should be analyzed separately, whereas the male–female dichotomy is equivocal. The three chi-square­ Dorsolateral Folds and Lateral Folds values comparing the male and female data are 0.612, 0.772, and 0.360. As it is improbable that realistically in- Application of the guidelines to the age and sex analy- creasing sample size would result in statistical significance ses results for both characters (Tables 12, 13) indicate that at the 5% level, the male and female data are combined age and sex categories are best combined for the intrasam- for the intrasample variation summary (Table 6) and for ple summary (Table 11) and in subsequent intersample subsequent intersample analyses. analyses.

Lip Stripe Male Secondary Sexual Characters

The variation in results of the age and sex analyses The distributions of states among the samples are pre- (Table 8) mirrors that described for the subocular spot. sented in Table 14. Juvenile and adult data are best analyzed separately. The male and female category data are equivocal. The single Measurement Data chi-­square test result for the Boca do Acre, Brazil data is 0.368. The distribution of occurrences of states from There is considerable variation in measurement value Cuzco Amazónico, Peru is almost identical and is rather ranges within samples (Table 15). The intrasample varia- similar for the Panama Canal sample. The male and fe- tion in all variables far exceeds the measurement error male data are not differentiated for the intrasample sum- variation. mary (Table 6) or in subsequent intersample analyses. There are unexpectedly large overlaps in body sizes (SVL) between specimens classified as juvenile and adult Belly Pattern males and females. For the four samples with adequate data for males, the overlap in SVL between the smallest The juvenile and adult analysis results (Table 9) are adult male and largest juvenile male ranges from 4.0 to equivocal as to whether the data are best analyzed com- 18.0 mm. For the two samples with adequate data for bined or separately. The three chi-square­ results involved females, the overlaps range from 8.8 to 22.0 mm. The are 0.348, 0.404, and 0.805. Realistically, increasing the largest values in overlap between juveniles and adults in 10 • smithsonian contributions to zoology

TABLE 6. Occurrence (percent) of dorsal pattern, suborbital spot, lip pattern, and belly pattern states within samples by age and region. Dorsal pattern states: A, uniform between interocular bar (if present) and sacrum; B, single chevron between interocular bar and sacrum without additional spots lateral to dorsal chevron; C, pattern complex with two dorsal chevrons and spotting all over dorsum, chevrons often elongate and fused with each other and interorbital bar. Suborbital spot states: A, distinct; B, indistinct; C, absent. Lip pattern states: A, noticeable broad light stripe extending from front of snout to commissural gland, bordered below by continuous or broken markings on upper lip and either bordered above by dark line under eye or not; B, narrow light stripe from at least nostril to lower pos- terior end of eye; C, narrow light stripe from below front of eye through commissural gland; D, lip region uniform (about same intensity as background dorsal color) with or without a dark stripe under the eye and/or dark bars/mottling on the upper lip. Belly pattern states: A, belly speckled, spotted, or mottled anteriorly only or with an anterior-posterior gradient; B, belly boldly mottled anteriorly only or with an anterior-posterior gradient; C, belly speckled or mottled with same intensity over entire belly; D, belly boldly mottled with same intensity over entire belly; E, belly without a pattern.

Dorsal pattern Suborbital spot Lip pattern Belly pattern

Juvenile/Adult State Juveniles Adults State Juveniles Adults State Juveniles Adults State data combined

Costa Rica/Panama Border (n = 45) (n = 27) (n = 45) (n = 27) (n = 45) (n = 27) (n = 72)

A 2 15 A 84 30 A 93 26 A 56 B 18 3 B 16 59 B 0 0 B 4 C 80 81 C 0 11 C 0 0 C 19 D 7 74 D 19 E 1

Panama Canal (n = 29) (n = 17) (n = 29) (n = 17) (n = 29) (n = 17) (n = 45)

A 10 41 A 69 47 A 52 24 A 58 B 10 29 B 24 41 B 0 0 B 4 C 80 29 C 7 12 C 3 0 C 20 D 45 76 D 7 E 11

Lower Río Chucunaque, Panama (n = 39) (n = 19) (n = 39) (n = 19) (n = 39) (n = 19) (n = 58)

A 5 37 A 64 90 A 51 37 A 50 B 13 0 B 31 10 B 0 0 B 0 C 82 63 C 5 0 C 8 0 C 45 C/D 5 0 D 2 D 36 63 E 3

Ilha do Maracá, Brazil (n = 28 (n = 8) (n = 28) (n = 8) (n = 28) (n = 8) (n = 36)

A 11 25 A 71 50 A 7 0 A 25 B 25 12 B 0 0 B 0 0 A/B 3 C 64 62 C 29 50 C 0 0 B 39 C/D 7 12 C 14 D 86 88 D 19 E 0

(continued) number 635 • 11

TABLE 6. (Continued)

Dorsal pattern Suborbital spot Lip pattern Belly pattern Juvenile/Adult State Juveniles Adults State Juveniles Adults State Juveniles Adults State data combined

Boca do Acre, Brazil (n = 12) (n = 24) (n = 12) (n = 24) (n = 12) (n = 24) (n = 36)

A 8 4 A 42 83 A 17 42 A 83 B 33 71 B 0 0 B 0 0 B 0 C 58 25 C 58 17 C 58 17 C 0 C/D 8 33 D 3 D 17 8 E 14

Cuzco Amazónico, Peru (n = 16) (n = 16) (n = 16) (n = 16) (n = 16) (n = 16) (n = 31)

A 6 31 A 88 62 A 93 50 A 90 B 19 44 B 12 31 B 0 0 B 0 C 75 25 C 0 6 C 7 25 C 10 D 0 25 D 0 E 0

TABLE 7. Age and sex intrasample analyses results for subocu- TABLE 8. Age and sex intrasample analyses results for lip stripe lar spot. Definitions: >OE, observer error threshold passed; NS, pattern. Definitions: >OE, observer error threshold passed; NS, chi-square results not statistically significant at≥ 0.05 level; NA, chi-square results not statistically significant at≥ 0.05 level; NA, data inadequate for chi-square analysis; c/w = compared with. data inadequate for chi-square analysis; c/w = compared with; A dash (–) indicates that it would be inappropriate to perform a a dash (–) indicates that it would be inappropriate to perform a significance test as the observer error threshold was not passed. significance test as the observer error threshold was not passed.

Juveniles c/w Adults Females c/w Males Juveniles c/w Adults Females c/w Males

Sample >OE Significance >OE Significance Sample >OE Significance >OE Significance

Lower Río Chucunaque, Yes NS Yes NS Lower Río Chucunaque, Yes NS No – Panama Panama Panama Canal Yes NS No – Panama Canal Yes NS Yes NA Costa Rica/Panama Border Yes 0.012 Yes NS Costa Rica/Panama Border Yes 0.0004 No – Cuzco Amazónico, Peru Yes NA Yes NA Cuzco Amazónico, Peru Yes NA Yes NA Boca do Acre, Brazil Yes NA Yes NA Boca do Acre, Brazil Yes NS Yes NS Ilha do Maracá, Brazil No – Yes NS Ilha do Maracá, Brazil No – No – this study noticeably exceed those from a study of the 11.0 mm; L. labyrinthicus juvenile-­adult male SVL over- Leptodactylus podicipinus–L. wagneri complex (Heyer, lap 22.4 mm; juvenile-adult­ female SVL overlap 11.0 1994:4–9). The overlaps are comparable for unpublished mm; L. savagei juvenile-­adult male SVL overlap 29.0 data for large species of the Leptodactylus pentadacty- mm.). These data suggest at least two nonexclusive expla- lus cluster (e.g., L. knudseni juvenile-adult­ male SVL nations: (1) the morphological proxies used to differenti- overlap 26.6 mm, juvenile-adult­ female SVL overlap ate adults from juveniles may not be consistently true and 12 • smithsonian contributions to zoology

TABLE 9. Age and sex intrasample analyses results for belly TABLE 10. Age and sex intrasample analyses results for thigh pattern. Definitions: >OE, observer error threshold passed; NS, pattern. Definitions: >OE, observer error threshold passed; NS, chi-square results not statistically significant at≥ 0.05 level; NA, chi-square results not statistically significant at≥ 0.05 level; NA, data inadequate for chi-square analysis. A dash (–) indicates that data inadequate for chi-square analysis; c/w = compared with. it would be inappropriate to perform a significance test as the A dash (–) indicates that it would be inappropriate to perform a observer error threshold was not passed; c/w = compared with. significance test as the observer error threshold was not passed.

Juveniles c/w Adults Females c/w Males Juveniles c/w Adults Females c/w Males

Sample >OE Significance >OE Significance Sample >OE Significance >OE Significance

Lower Río Chucunaque, No – No – Lower Río Chucunaque, No – No – Panama Panama Panama Canal Yes NS Yes NS Panama Canal No – No – Costa Rica/Panama Border Yes – No – Costa Rica/Panama Border No – Yes NA Cuzco Amazónico, Peru No NS No – Cuzco Amazónico, Peru No – No – Boca do Acre, Brazil No NS No – Boca do Acre, Brazil No – No – Ilha do Maracá, Brazil Yes – No – Ilha do Maracá, Brazil Yes NA No –

(2) the variation observed in Leptodactylus is related to samples were used because the purpose of this evaluation overall adult size. is to determine whether the characters demonstrate any patterns of geographic variation. There are three major geographic areas occupied by the Leptodactylus bolivi- INTERSAMPLE VARIATION anus complex: (1) Middle America, for which three main- land samples and two island samples, one Atlantic and After mapping localities, 20 regional samples were cho- one Pacific were selected; (2) populations predominantly sen to analyze the nature of variation among them (Figure occurring in South American river basins draining to the 2). The sampling strategy was to maximize both geographic Caribbean for which eight samples were selected (the coverage and sample sizes. This strategy was not possible in Northern Roraima sample is geographically near the rest some areas of coverage, and smaller than desired samples of the samples but is in an Amazonian drainage area); and are included because of their importance in geographic rep- (3) populations occurring in river basins draining to the resentation or potential nomenclatural impact. Amazon River, for which six samples were selected. Geo- graphically proximate samples are compared among these Patterns and Folds three major geographic areas to determine overall patterns of differentiation. The approach used to analyze intersample variation The full set of analytic results are shown for the first for the morphological characters involving patterns and character analyzed as an example of the comparisons that folds follows that used in the analysis of intrasample varia- exceed observer error, those for which χ2 analyses are inap- tion. Any comparisons between samples are first analyzed propriate (only one state with an expected sample size of 5 to determine whether the amount of variation exceeds or greater), and the level and variation of χ2 values when χ2 observer error. For the comparisons that exceed observer values were appropriate. Results for the other characters error, chi-­square tests were used when appropriate to de- are available from WRH on request and will be placed in termine statistical significance. Juvenile and adult data are the Smithsonian Institution Archives by 1 January 2012. analyzed separately for dorsal pattern, subocular spot, and lip stripe characters. Juvenile and adult data are combined Dorsal Pattern for analysis of belly pattern, thigh pattern, shank pattern, dorsolateral fold, and lateral fold characters. There are no significant differences among the three Rather than make comparisons between all pairs of mainland Middle American samples (Table 16). The Gulf samples for the characters involved, three networks of of Panama sample does not differ from the Panama Canal number 635 • 13

TABLE 11. Occurrence (percent) of thigh pattern, shank pattern, dorsolateral fold, and lateral fold states within region samples. Thigh pattern states: A, posterior thigh with a uniform or variable dispersion of melanophores (weakly to moderately mottled) usually for entire extent of posterior thigh; B, posterior thigh boldly mottled with large light irregular spots/vermiculations on a dark background to the light marks being more extensive than the dark areas, the light markings sometimes coalesced. Shank pattern states: A, uniform; B, short dark transverse bars extending less than halfway across dorsal surface of shank; C, at least one dark transverse bar extending halfway or more across the dorsal surface of the shank. Dorsolateral folds states: A, dorsolateral fold distinct and continuous from supratympanic fold to end of body, or with a brief interruption of the fold, or the fold distinct only from the supratympanic fold to the sacrum; B, dorsolateral fold with several interruptions to indescernable (including lack of fold due to poor preservation). Lateral folds states: A, lateral fold distinct or slightly interrupted from supratympanic fold to leg; B, lateral fold distinct only in groin region, indis- tinct overall, or indiscernible (including due to poor preservation).

Thigh pattern Shank pattern Dorsolateral folds Lateral folds State Occurrence (%) State Occurrence (%) State Occurrence (%) State Occurrence (%)

Costa Rica/Panama Border (n = 72) (n = 72) (n = 72) (n = 72)

A 17 A 7 A 71 A 29 A/B* 1 B 28 B 29 B 71 B 81 C 65 Panama Canal (n = 46) (n = 46) (n = 45) (n = 45)

A 0 A 2 A 67 A 29 A/B* 6 B 50 B 33 B 71 B 93 B/C* 2 C 46 Lower Río Chucunaque, Panama (n = 58) (n = 58) (n = 58) (n = 58)

A 0 A 0 A 81 A 57 A/B* 3 B 24 B 19 B 43 B 97 C 76 Ilha do Maracá, Brazil (n = 36) (n = 36) (n = 36) (n = 36)

A 8 A 0 A 100 A 0 A/B* 14 B 42 B 0 B 100 B 78 B/C* 3 C 56 Boca do Acre, Brazil (n = 36) (n = 36) (n = 36) (n = 33)

A 100 A 3 A 97 A 30 B 0 B 78 B 3 B 70 C 19 Cuzco Amazónico, Peru (n = 32) (n = 32) (n = 32) (n = 32)

A 100 A 0 A 100 A 12 B 0 B 38 B 0 B 88 C 62

*Indicates a condition that is intermediate between the two states. 14 • smithsonian contributions to zoology

TABLE 12. Age and sex intrasample analyses results for dorsolat- TABLE 13. Age and sex intrasample analyses results for lateral eral folds. Definitions: >OE, observer error threshold passed; NS, folds. Definitions: >OE, observer error threshold passed; NS, chi-square results not statistically significant at ≥0.05 level; NA, chi-square results not statistically significant at≥ 0.05 level; NA, data inadequate for chi-square analysis; c/w = compared with. data inadequate for chi-square analysis; c/w = compared with. A dash (–) indicates that it would be inappropriate to perform a A dash (–) indicates that it would be inappropriate to perform a significance test as the observer error threshold was not passed. significance test as the observer error threshold was not passed.

Juveniles c/w Adults Females c/w Males Juveniles c/w Adults Females c/w Males

Sample >OE Significance >OE Significance Sample >OE Significance >OE Significance

Lower Río Chucunaque, Yes NS No – Lower Río Chucunaque, Yes NS No – Panama Panama Panama Canal No – Yes NS Panama Canal No – Yes NS Costa Rica/Panama Border Yes NS No – Costa Rica/Panama Border No – No – Cuzco Amazónico, Peru No – No – Cuzco Amazónico, Peru No – Yes NA Boca do Acre, Brazil No – No – Boca do Acre, Brazil No – No – Ilha do Maracá, Brazil No – No – Ilha do Maracá, Brazil No – No –

sample but does differ from the other two mainland sam- There are no statistically significant differences among ples. The (Pacific) Gulf of Panama sample does not differ the Amazonian drainage network of samples (Table 16). from the (Caribbean) San Andrés/Providencia sample. The San Andrés/Providencia sample additionally does not dif- Subocular Spot fer from the Panama Canal sample, but does differ from the Southern Costa Rica/Eastern Panama and Darién There are no demonstrable differences among juve- samples. niles in the Middle American network samples. The adult The geographically adjacent Darién and Northern data show differences between the mainland samples Magdalena Drainage samples do not differ from each Southern Costa Rica/Eastern Panama compared with the other (Table 16). Darién data (χ2 = 0.001), and the Panama Canal sample Among the primarily Caribbean drainage samples, compared with the Darién data (χ2 = 0.020). The adult there are only two statistically significant results: juvenile data also differ between the San Andrés/Providencia sam- Northern Guyana data compared with juvenile Suriname ple and the Southern Costa Rica/Eastern Panama sample data (Table 16 sample comparisons K:M) and juvenile (χ2 = 0.035) and with the Darién sample (χ2 = 0.001). Northern Guyana data compared with juvenile Northern The adult data differ between the samples that geo- Roraima data (Table 16 sample comparisons K:N). As graphically link the Darién sample with the Northern there are 72 comparisons within this network of samples, Magdalena Drainage sample (χ2 = 0.044). Juvenile data having two statistically significant results is well within are distinct for the following comparisons: Middle-­ statistical expectations. Southern Magdalena Drainage sample compared with the There are no statistically significant differences be- Northern Roraima Sample (χ2 = 0.033); Meta Drainage tween the Meta Drainage and the Amazonian drainage sample compared with the Northern Roraima sample (χ2 samples (Rio Solimões/Amazonas, Río Ucayali, Madre de = 0.006); Southern Guyana sample compared with the Dios, Lower Rio Purús/Rio Acre, Middle/Lower Río Ma- Suriname sample (χ2 = 0.022); and the Southern Guyana deira, South Central Bolivia) (Table 16). There is only one sample compared with the Northern Roraima sample (χ2 = statistically significant difference in the juvenile data for 0.004). Adult data are distinct for the following compari- the Northern Roraima–South Central Bolivia compari- sons: Northern Magdalena Drainage sample compared son (Table 16, sample comparisons N:T). There are no with the Meta Drainage sample (χ2 = 0.042), Meta Drain- compelling data to favor the Meta Drainage sample or the age sample with the Southern Guyana sample (χ2 = 0.003), Northern Roraima sample as showing a pattern in com- Meta Drainage sample with the Suriname sample (χ2 = mon with the rest of the Amazonian drainage samples. 0.009), and the Meta Drainage sample with the Northern number 635 • 15

TABLE 14. Male secondary sexual character state occurrences (percent) within samples. Arm states: A, arm not hypertrophied; B, arm hypertrophied. Lower jaw tubercle states: A, absent; B, tuberculate. Chest states: A, no patch of tubercles; B, central patch of tubercles; C, lateral patches of tubercles in addition to central patch. Thumb spine states: A, one modest chisel-shaped spine; B, one hypertrophied chisel-shaped spine; C, two round spines. Tympanum states: A, annulus lacking tubercles; B, annulus tuberculate. A dash (–) indicates no state or no percent occurrence for samples with n = 0.

Arm hypertrophy Jaw tubercles Chest tubercles Thumb spines Tympanum tubercles State Value State Value State Value State Value State Value

Costa Rica–Panama Border n = 15 n = 12 n = 15 n = 15 n = 15

A 33 A 67 A 53 A 0 A 0 B 67 B 33 B 47 B 0 B 100 C 0 C 100

Panama Canal n = 11 n = 10 n = 11 n = 11 n = 11

A 64 A 80 A 64 A 0 A 0 B 36 B 20 B 36 B 0 B 100 C 0 C 100

Lower Río Chucunaque, Panama n = 8 n = 7 n = 8 n = 8 n = 8

A 50 A 57 A 38 A 0 A 25 B 50 B 43 B 62 B 0 B 75 C 0 C 100

Ilha do Maracá, Brazil n = 1 n = 0 n = 1 n = 1 n = 0

A 100 – – A 100 A 100 – – B 0 – – B 0 B 0 – – C 0 C 0

Boca do Acre, Brazil n = 4 n = 2 n = 4 n = 4 n = 4

A 75 A 50 A 100 A 25 A 0 B 25 B 50 B 0 B 75 B 100 C 0 C 0

Cuzco Amazónico, Peru n = 5 n = 4 n = 5 n = 5 n = 4

A 0 A 25 A 20 A 0 A 0 B 100 B 75 B 20 B 100 B 100 C 60 C 0

Roraima sample (χ2 = 0.027). Overall, the Meta Drainage sample (adult data, χ2 = 0.006) and the Madre de Dios sample stands out as being distinctive within this network sample (adult data, χ2 = 0.004). The Northern Roraima of samples. sample differs from the Madre de Dios sample (juvenile The geographically network-linking­ Meta Drainage data χ2 = 0.011), the Lower Purús/Rio Amazonas sample sample differs from both the Rio Solimões/Amazonas (adult data, χ2 = 0.003), and the South Central Bolivia 16 • smithsonian contributions to zoology

TABLE 15. Intrasample variation in measurement data. Definitions: n, sample size; A, Lower Río Chucunaque, Panama sample; B, Panama Canal sample; C, Costa Rica–Panama Border sample; D, Cuzco Amazónico, Peru sample; E, Boca do Acre, Brazil sample; F, Ilha do Maracá, Brazil sample; Min, minimum value (mm); Max, maximum value (mm); Range, range of values (mm); SVL, snout–vent length; HL, head length; HW, head width; EN, eye–nostril distance; TD, tympanum diameter. A dash (–) indicates there is only one individual in the sample.

Males Females

Variable Sample n Min Max Range n Min Max Range

SVL A 9 78.9 97.9 19.0 16 75.5 89.3 13.8 B 11 67.8 101.3 33.5 6 79.0 95.5 16.5 C 15 76.0 92.8 16.8 12 74.5 86.8 12.3 D 5 106.2 116.0 9.8 11 77.9 99.8 21.9 E 4 79.0 97.4 18.4 20 63.0 86.3 23.3 F 1 – 86.0 – 7 66.0 95.5 29.5 HL A 9 29.1 36.2 7.1 16 28.4 32.0 3.6 B 11 25.2 34.7 9.5 6 28.1 35.1 7.0 C 15 28.4 33.3 4.9 11 26.4 32.4 6.0 D 5 36.6 40.4 3.8 11 29.3 34.4 5.1 E 4 28.7 37.3 8.6 20 24.0 32.7 8.7 F 1 – 32.5 – 7 25.5 34.0 8.5 HW A 9 27.1 36.9 9.8 16 25.0 29.2 4.2 B 11 23.3 36.5 13.2 6 26.3 32.7 6.4 C 15 26.0 33.1 7.1 12 25.0 28.6 3.6 D 5 35.2 40.4 5.2 11 23.8 31.5 7.7 E 4 24.2 31.0 6.8 20 20.0 28.1 8.1 F 1 – 29.5 – 7 22.0 31.0 9.0 EN A 9 8.5 9.5 1.0 16 7.7 9.4 1.7 B 11 7.4 9.9 2.5 6 7.8 9.8 2.0 C 15 8.0 9.8 1.8 12 7.5 9.0 1.5 D 5 11.0 11.7 0.7 11 8.3 9.7 1.4 E 4 7.8 9.8 2.0 20 6.5 8.7 2.2 F 1 – 9.0 – 7 7.0 9.0 2.0 TD A 9 5.5 7.1 1.6 16 5.2 6.2 1.0 B 11 5.0 7.8 2.8 6 5.6 6.6 1.0 C 15 5.2 6.4 1.2 12 5.2 6.3 1.1 D 5 6.8 8.5 1.7 11 6.0 7.2 1.2 E 4 5.8 7.6 1.8 20 4.5 6.5 2.0 F 1 – 6.0 – 7 5.0 6.5 1.5 Thigh A 9 37.3 43.0 5.7 15 32.3 41.0 8.7 B 11 30.4 48.7 18.3 6 36.3 46.7 10.4 C 15 33.8 42.0 8.2 12 30.9 39.2 8.3 D 5 42.8 56.7 13.9 11 36.1 45.0 8.9 E 4 37.8 47.1 9.3 20 30.0 39.8 9.8 F 1 – 46.5 – 7 30.5 46.0 15.5 Shank A 9 40.8 46.7 5.9 15 37.2 44.3 7.1 B 11 34.8 49.0 14.2 6 41.3 48.9 7.6 C 15 38.1 45.6 7.5 12 36.6 43.5 6.9 D 5 51.6 58.5 6.9 11 40.6 48.2 7.6 E 4 41.5 51.3 9.8 20 33.5 47.5 14.0 F 1 – 46.5 – 7 36.5 50.0 13.5 Foot A 9 39.8 46.6 6.8 16 38.1 44.4 6.3 B 11 37.2 48.8 11.6 6 40.2 50.2 10.0 C 15 37.8 45.5 7.7 12 37.6 44.7 7.1 D 5 53.5 60.3 6.8 11 42.8 50.2 7.4 E 4 43.5 52.4 8.9 20 35.0 46.6 11.6 F 1 – 48.0 – 7 37.0 49.0 12.0 number 635 • 17

E H B C J F A D I K M G N L

O S P R

Q

T

FIGURE 2. Map showing locations of 20 regional samples. A, Costa Rica/Panama Border; B, Panama Canal; C, Darién; D, Gulf of Panama; E, San Andrés/Providencia; F, Northern Magdalena Drainage; G, Middle/Southern Magdalena Drainage; H, Santa Marta; I, Meta Drainage; J, Caracas; K, Northern Guyana; L, Southern Guyana; M, Suriname; N, Northern Roraima; O, Rio Solimões/Amazonas; P, Río Ucuyali; Q, Madre de Dios; R, Lower Rio Purus/Rio Acre; S, Middle/Lower Rio Madeira; and T, South Central Bolivia.

sample (juvenile data, χ2 = 0.001). Both the Meta Drainage Lip Stripe and Northern Roraima samples are weakly differentiated (at best) from the remaining Amazonian drainage samples. There are only two statistically significant different Within the Amazonian drainage network exclusive of comparisons within the Middle American network of the Northern Roraima sample, the following comparisons samples: Southern Costa Rica/Eastern Panama with the are distinct: Rio Solimões/Rio Amazonas with the Lower Panama Canal samples (juvenile data, χ2 = 0.016) and the Rio Purús/Rio Acre samples (adult data, χ2 = 0.012); Southern Costa Rica/Eastern Panama with the Darién Madre de Dios with the South Central Bolivia samples samples (juvenile data, χ2 = 0.010). (juvenile data, χ2 = 0.003); and the Lower Rio Purús/Rio The data do not statistically differ between the Acre with the South Central Bolivia samples (adult data, network-­linking Darién and Northern Magdalena Drain- χ2 = 0.032). age samples. 18 • smithsonian contributions to zoology

TABLE 16. Intersample comparisons for dorsal pattern data. Significant chi-square χ( 2) values are in italic font and followed by an as- terisk. Sample definitions: A, Southern Costa Rica/Eastern Panama; B, Panama Canal; C, Darién; D, Gulf of Panama; E, San Andrés/ Providencia; F, Northern Magdalena Drainage; G, Middle/Southern Magdalena Drainage; H, Santa Marta; I, Meta Drainage; J, Cara- cas; K, Northern Guyana; L, Southern Guyana; M, Suriname; N, Northern Roraima; O, Rio Solimões/Amazonas; P, Río Ucayali; Q, Madre de Dios; R, Lower Rio Purús/Rio Acre; S, Middle/Lower Rio Madeira; T, South Central Bolivia. Other abbreviations: n, sample size; NA, samples not appropriate for chi-square analysis. A dash (–) indicates that it was inappropriate to perform a significance test as the observer error threshold was not passed.

Juvenile data Adult data Juvenile data Adult data

Sample Exceeds Exceeds Sample Exceeds Exceeds pair observer observer pair observer observer comparisons n error χ2 n error χ2 comparisons n error χ2 n error χ2

A–B 49, 31 No – 29, 21 Yes 0.111 J–K 30, 22 Yes 0.486 27, 14 Yes 0.235 A–C 49, 79 No – 29, 49 Yes 0.102 J–L 30, 39 No – 27, 18 Yes 0.214 B–C 31, 79 No – 21, 49 Yes 0.343 J–M 30, 15 Yes 0.541 27, 13 Yes 0.628 A–D 49, 2 Yes 0.691 49, 21 Yes 0.021* J–N 30, 46 No – 27, 23 Yes 0.614 B–D 31, 2 Yes 0.324 21, 21 Yes 0.296 K–L 22, 39 Yes 0.155 14, 18 Yes 0.434 C–D 79, 2 Yes 0.793 49, 21 Yes 0.029* K–M 22, 15 Yes 0.042* 14, 13 Yes 0.248 D–E 2, 7 Yes NA 21, 29 Yes 0.292 K–N 22, 46 Yes 0.036* 14, 23 Yes 0.700 A–E 49, 7 Yes NA 29, 29 Yes 0.007* L–M 39, 15 Yes 0.619 18, 13 No – B–E 31, 7 Yes NA 21, 29 No – L–N 39, 46 No – 18, 23 Yes 0.646 C–E 79, 7 Yes NA 49, 29 Yes 0.009* M–N 15, 46 Yes 0.481 13, 23 Yes 0.754 C–F 79, 20 Yes 0.390 49, 23 No – I–O 6, 9 Yes NA 12, 9 Yes NA F–G 20, 75 No – 23, 85 No – I–P 6, 18 No – 12, 8 Yes NA F–H 20, 64 No – 23, 45 Yes 0.313 I–Q 6, 73 No – 12, 19 Yes 0.857 F–I 20, 6 Yes NA 23, 12 Yes 0.638 I–R 6, 13 No – 12, 29 Yes 0.260 F–J 20, 30 No – 23, 27 No – I–S 6, 2 Yes NA 12, 8 Yes NA F–K 20, 22 Yes 0.338 23, 14 Yes 0.497 I–T 6, 51 Yes NA 12, 7 Yes NA F–L 20, 39 No – 23, 18 Yes 0.497 N–O 46, 9 Yes 0.582 23, 9 Yes 0.662 F–M 20, 15 Yes 0.222 23, 13 Yes 0.781 N–P 46, 18 No – 23, 8 Yes 0.597 F–N 20, 46 No – 23, 23 No – N–Q 46, 73 Yes 0.092 23, 19 No – G–H 75, 64 No – 23, 45 Yes 0.411 N–R 46, 13 No – 23, 29 Yes 0.098 G–I 75, 6 Yes 0.978 23, 12 Yes 0.792 N–S 46, 2 Yes 1.000 23, 8 Yes 0.814 G–J 75, 30 No – 23, 27 No – N–T 46, 51 Yes 0.033* 23, 7 Yes 0.875 G–K 75, 22 Yes 0.186 23, 14 Yes 0.663 O–P 9, 18 Yes NA 9, 8 Yes NA G–L 75, 39 No – 23, 18 Yes 0.507 O–Q 9, 73 Yes 0.356 9, 19 Yes 0.706 G–M 75, 15 No – 23, 13 Yes 0.838 O–R 9, 13 Yes 0.545 9, 29 Yes 0.600 G–N 75, 46 No – 23, 23 Yes 0.866 O–S 9, 2 No – 9, 8 Yes NA H–I 20, 6 Yes 0.963 23, 12 Yes 0.524 O–T 9, 51 Yes 0.304 9, 7 Yes NA H–J 20, 30 No – 23, 27 Yes 0.556 P–Q 18, 73 No – 8, 19 Yes 0.414 H–K 20, 22 Yes 0.174 23, 14 Yes 0.505 P–R 18, 13 No – 8, 29 Yes 0.226 H–L 20, 39 No – 23, 18 Yes 0.152 P–S 18, 2 Yes 1.000 8, 8 Yes NA H–M 20, 15 No – 23, 13 Yes 0.451 P–T 18, 51 Yes 0.221 8, 7 Yes NA H–N 20, 46 No – 23, 23 Yes 0.630 Q–R 73, 13 Yes 0.483 19, 29 Yes 0.475 I–J 6, 30 Yes 0.915 12, 27 Yes 0.343 Q–S 73, 2 Yes 0.743 19, 8 No – I–K 6, 22 Yes 0.758 12, 14 No – Q–T 73, 51 No – 19, 7 No – I–L 6, 39 No – 12, 18 Yes 0.480 R–S 13, 2 Yes NA 29, 8 No – I–M 6, 15 Yes NA 12, 13 Yes 0.352 R–T 13, 51 Yes 0.304 29, 7 Yes 0.814 I–N 6, 46 No – 12, 23 No – S–T 2, 51 Yes 0.637 8, 7 No – ­ number 635 • 19

The data statistically differ between several of the pri- Guyana (χ2 = 0.002), Northern Magdalena Drainage marily Caribbean draining network samples: Northern with Southern Guyana (χ2 = 0.001), Northern Magdalena Magdalena Drainage with Southern Guyana (adult data, Drainage with Northern Roraima (χ2 < 0.001), Middle/ χ2 = 0.018); Northern Magdalena Drainage with Suriname Southern Magdalena Drainage with Northern Guyana (adult data, χ2 = 0.018); Middle/Southern Magdalena (χ2 = 0.004), Middle/Southern Magdalena Drainage with Drainage with Northern Guyana (adult data, χ2 = 0. 018); Southern Guyana (χ2 < 0.001), Middle/Southern Magda- Santa Marta with Northern Guyana (juvenile data, χ2 < lena Drainage with Northern Roraima (χ2 < 0.001), Santa 0.001, adult data, χ2 = 0.001); Santa Marta with Southern Marta with Northern Guyana (χ2 = 0.010), Santa Marta Guyana (juvenile data, χ2 = 0.003, adult data, χ2 = 0.025); with Southern Guyana (χ2 = 0.001), Santa Marta with Santa Marta with Suriname (adult data, χ2 = 0.025); Meta Northern Roraima (χ2 < 0.001), and Caracas with South- Drainage with Northern Guyana (adult data, χ2 = 0.002); ern Guyana (χ2 = 0.030). In general, there is a contrast Meta Drainage with Southern Guyana (adult data, χ2 = between coastal and river valley samples from Colombia 0.035); Caracas with Northern Guyana (juvenile data, and Venezuela with the Guyana and Northern Roraima χ2 = 0.009; adult data, χ2 = 0.012); Northern Guyana with samples. Southern Guyana (adult data, χ2 = 0.035); Northern Guy- There are no statistically significant differences be- ana with Northern Roraima (juvenile data, χ2 < 0.001; tween the network-­linking Meta Drainage sample with the adult data, χ2 = 0.001); Southern Guyana with Northern Amazonian network samples. In contrast, there are three Roraima (juvenile data, χ2 < 0.001), and Suriname with statistically significant differences among the Northern Northern Roraima (adult data, χ2 = 0.053). The amount Roraima and other Amazonian network samples: Madre of demonstrable variation in this network clearly exceeds de Dios (χ2 < 0.001), Lower Rio Purús/Rio Acre (χ2 = any variation that could be accounted for by chance alone. 0.001), and South Central Bolivia (χ2 < 0.001). The network-­linking Meta Drainage sample differs There are no statistically significant differences among from the Madre de Dios sample (adult data, χ2 = 0.052) the primarily Amazonian network samples. and the Lower Rio Purús/Rio Acre sample (juvenile data, χ2 = 0.018; adult data, χ2 = 0.021). The Northern Roraima Thigh Pattern sample differs from the Rio Solimões/Rio Amazonas sam- ple (adult data, χ2 = 0.031), Madre de Dios (juvenile data, There are no significant differences among the Middle χ2 < 0.001; adult data, χ2 = 0.035), Lower Rio Purús/Rio American samples or between the network-­linking Darién Acre (juvenile data, χ2 = 0.054; adult data, χ2 < 0.001), and with the Northern Magdalena Drainage samples. South Central Bolivia (juvenile data, χ2 < 0.001) samples. There are statistically significant differences between There seems to be more differentiation between the North- several of the primarily Caribbean network samples: ern Roraima sample with the other Amazonian drainage Northern Magdalena Drainage with Meta Drainage (χ2 = samples than between the Meta Drainage sample and the 0.003), Northern Magdalena Drainage with Southern primarily Amazonian network samples. Guyana (χ2 < 0.001), Middle/Southern Magdalena Drain- Within the primarily Amazonian drainage sample net- age with Southern Guyana (χ2 < 0.001), Santa Marta with work, there are only two statistically significant compari- Meta Drainage (χ2 = 0.034), Santa Marta with Southern sons: the Río Ucayali Drainage sample with the Lower Guyana (χ2 < 0.001), Meta Drainage with Caracas (χ2 = Rio Purús/Rio Acre sample (juvenile data, χ2 = 0.010), and 0.005), Meta Drainage with Northern Guyana (χ2 = the Lower Rio Purús/Rio Acre sample with the South Cen- 0.016), Meta Drainage with Suriname (χ2 = 0.015), Ca- tral Bolivia sample (juvenile data, χ2 = 0.040). racas with Southern Guyana (χ2 < 0.001), Northern Guy- ana with Southern Guyana (χ2 < 0.001), Southern Guyana Belly Pattern with Suriname (χ2 < 0.001), Southern Guyana with North- ern Roraima (χ2 < 0.001). There are no significant differences among the Middle Each comparison between the network linking Meta American samples or between the network-­linking Darién Drainage and Northern Roraima samples is statistically with the Northern Magdalena Drainage samples. significant with the primarily Amazonian network sam- There are several statistically significant different com- ples: Meta Drainage with Rio Solimões/Rio Amazonas parisons within the primarily Caribbean drainage network (χ2 = 0.027), Río Ucayali (χ2 = 0.011), Madre de Dios (χ2 = samples: Northern Magdalena Drainage with Northern 0.001), Lower Rio Purús/Rio Acre (χ2 = 0.001), Middle/ 20 • smithsonian contributions to zoology

FIGURE 3. Variation in posterior thigh patterns among 20 regional samples. Pattern A, light gray; pattern B, white; intermediate pattern A/B, dark gray.

Lower Rio Madeira (χ2 = 0.004), and South Central Bo- Shank Pattern livia (χ2 = 0.001) samples; Northern Roraima with Rio Solimões/Rio Amazonas (χ2 = 0.008), Rio Ucayali (χ2 < There are no statistically significant differences among 0.001), Madre de Dios (χ2 < 0.001), Lower Rio Purús/Rio the Middle American samples or between the network-­ Acre (χ2 < 0.001), Middle/Lower Rio Madeira (χ2 = 0.048), linking Darién with the Northern Magdalena Drainage and South Central Bolivia (χ2 < 0.001) samples. samples. Within the primarily Amazonian network samples There are four statistically significant differences only the Río Ucayali and Madre de Dios samples differ among the primarily Caribbean drainage network samples: significantly χ( 2 = 0.022). Northern Magdalena Drainage with Northern Guyana The thigh pattern variation demonstrates the most (χ2 = 0.009), Meta Drainage with Northern Guyana (χ2 = geographically coherent pattern (Figure 3). 0.017), Caracas with Northern Guyana (χ2 = 0.004), and number 635 • 21

Northern Guyana with Northern Roraima (χ2 = 0.003). The Magdalena Drainage with Northern Roraima (χ2 = 0.013), Northern Guyana sample stands out within this network. Santa Marta with Northern Guyana (χ2 = 0.023), Santa There is one statistically significant difference be- Marta with Northern Roraima (χ2 < 0.001), Caracas with tween the network-­linking Meta Drainage sample with Northern Guyana (χ2 = 0.009), and Caracas with North- the Lower Rio Purús/Rio Acre sample (χ2 = 0.011). There ern Roraima (χ2 = 0.003). are no statistically significant differences between the The network-­linking Meta Drainage sample statis- network-­linking Roraima sample with the other Amazo- tically differs from the Río Ucayali sample (χ2 = 0.025). nian samples. The network-­linking Northern Roraima sample statisti- There are two statistically significant differences cally differs from the Río Ucayali (χ2 = 0.007), Lower Rio among the primarily Amazonian drainage network sam- Purús/Rio Acre (χ2 = 0.048), and South Central Bolivia ples: Madre de Dios with Lower Rio Purús/Rio Acre (χ2 = (χ2 = 0.013) samples. 0.005) and Lower Rio Purús/Rio Acre with South Central There are three sample comparisons that statistically Bolivia (χ2 = 0.016). differ significantly within the Amazonian network: Río Ucayali with Madre de Dios (χ2 = 0.005), Madre de Dios Dorsolateral Folds with Lower Rio Purús/Rio Acre (χ2 = 0.030), and Madre de Dios with South Central Bolivia (χ2 = 0.005). There are no statistically significant differences among the Middle American samples or between the network-­ Male Secondary Sexual Characters linking Darién with the Northern Magdalena Drainage samples. Sample sizes are inadequate to perform statistical There are several statistically significant differences tests for many of the sample comparisons. In addition, the among the primarily Caribbean drainage network sam- problem of seasonal development of sexual characters has ples: Northern Magdalena Drainage with Santa Marta to be addressed. Comparisons among samples should be (χ2 = 0.010), Northern Magdalena Drainage with Caracas made on males that are sexually active and have devel- (χ2 < 0.001), Northern Magdalena Drainage with South- oped their full extent of secondary sexual characteristics. ern Guyana (χ2 = 0.002), Northern Magdalena Drainage For this purpose, the presence of vocal slits is inappropri- with Suriname (χ2 = 0.023), Northern Magdalena Drain- ate, as once the slits are developed, they are not lost and age with Northern Roraima (χ2 < 0.001), Middle/­Southern regained on a seasonal basis. For comparative purposes, Magdalena Drainage with Northern Roraima (χ2 = 0.013), any noticeable arm hypertrophy is used as a proxy for full and Northern Guyana with Northern Roraima (χ2 = development of secondary sexual characters, realizing that 0.033). The distinctiveness of the Northern Magdalena this proxy is approximate and not definitive. Drainage sample is due to more occurrences of the dor- The data are best analyzed visually with the raw num- solateral folds with several interruptions to indiscernible bers of state occurrences for all 20 geographic samples conditions: it is possible that this is due to poor preserva- (Table 17). tion rather than a real sample difference. There are two obvious features in the data. First, there There are no statistically significant differences be- is no coherent pattern to the state occurrences for the jaw tween the network-linking­ Meta Drainage and Northern tubercle and tympanic annulus tubercle characters. The Roraima samples with the rest of the Amazonian drainage most reasonable interpretations for these two characters network samples. are that (1) tympanic annulus tubercles, once developed, There are no statistically significant differences be- are not shed seasonally; and (2) jaw tubercles are shed sea- tween any of the Amazonian network samples. sonally and are redeveloped each year by all adult males in all populations. Second, there is geographically coher- Lateral Folds ent variation among the states of the chest tubercles and thumb spine characters. As can be seen from the raw data There are no statistically significant differences among (Table 17), several of the samples have individuals with the Middle American samples or between the network-­ only a central patch of chest tubercles (state A) as well as linking Darién with the Northern Magdalena Drainage individuals with central and lateral patches of chest tu- samples. bercles (state B). This variation is interpreted to be onto- The following primarily Caribbean draining network genetic; that is, for individuals that have both central and samples differ statistically significantly: Middle/Southern lateral tubercle patches, the central patch develops before 22 • smithsonian contributions to zoology

lateral patches (Rio Purús/Rio Acre and Lower/Middle TABLE 17. Number of males exhibiting secondary sexual char- Madeira samples). The only adult male in the Río Ucayali acter states. Number of occurrences may not be equal among sample did not have noticeably hypertrophied arms. The characters due to missing observations. Samples are defined geographically most incongruous sample is the Suriname in Table 16. Jaw tubercle states: A, absent; B, present. Chest sample, where all three males with chest tubercles have ­tubercle states: A, absent; B, central patch; C, central and lat- eral patches. Thumb spine states: A, 1 slightly chisel-shaped; B, only central patches. 1 robustly chisel-shaped; C, 2 round. Tympanic tubercle states: Variation in thumb spines indicates three groupings A, absent; B, present. A dash (–) indicates no specimens had the of the geographic samples. The first is the contrast of two character state involved. versus one spine, with the Costa Rica/Eastern Panama through Caracas samples having two spines (see Figure 13A) and all other samples characterized by having one Jaw Chest Thumb Tympanic spine. There is also variation among the one spine sam- tubercles tubercles spines tubercles ples whether the spine is slightly chisel shaped (see Fig- Sample A B A B C A A/B* B C A B ure 13C) or obviously chisel shaped (see Figure 13B). The

A 3 5 2 8 0 0 0 0 10 0 10 Río Ucayali sample male lacking arm hypertrophy has a B 1 5 1 6 0 0 0 0 7 0 7 medium-­sized slightly chisel-shaped­ spine, suggesting that C 0 5 0 10 0 0 0 0 10 0 10 the spine is not fully developed. D 2 3 1 6 0 0 0 0 7 0 7 E 2 10 4 13 0 0 0 0 16 0 14 Measurement Data F 3 1 3 6 0 0 0 0 9 1 8 G 6 26 2 36 0 0 0 0 42 1 41 Discriminant function analyses were performed on the H 1 3 4 9 0 0 0 0 13 0 13 20 geographic samples separately for the female and male I – – 1 3 0 0 0 0 3 0 3 data. The data sets were run with SYSTAT 11 software J 2 3 2 6 0 0 0 0 8 0 8 (Engelman et al., 2004) using the complete, forward, and K 4 3 4 1 2 4 3 0 0 0 7 backward stepwise models. The results for the complete, L 1 0 1 0 1 2 0 0 0 0 2 forward, and backward stepwise models are very similar; M 1 3 3 3 0 4 2 0 0 1 5 N – – 0 1 3 4 0 0 0 0 4 only the complete model results are discussed. O 0 1 0 0 1 0 0 1 0 0 1 The first canonical vector accounts for 63% of the P – – – – – – – – – – – cumulative dispersion for the female data, the first two Q 1 3 3 0 2 0 0 6 0 0 5 canonical vectors account for 81%, and 90% is reached R – – 1 0 0 0 0 1 0 0 1 with the third canonical vector. None of the 20 samples S – – 2 0 0 0 0 2 0 0 2 are completely correctly identified in the posterior clas- T 0 2 1 0 1 0 0 2 0 0 2 sification matrix using all data nor in the jackknife clas- sification matrix. The poor discrimination of the data set *In the case of thumb spines, there were some individuals that exhibited intermediate states between state A and state B. is reflected in the plot of the first two canonical vectors (Figure 4). Three distinctive sample groups can be dis- tinguished from each other and the rest of the samples: the lateral patches develop on an annual basis. This inter- (1) the San Andrés/Providencia and Northern Magdalena pretation is supported by comparison of the jaw tubercle Drainage samples, (2) the Gulf of Panama sample, and (3) and chest tubercle data, assuming that both jaw and chest the Southern Guyana sample (Figure 4). tubercles start developing simultaneously each season in The first canonical vector accounts for 55% of the cu- a given male. Thus the critical aspect for this character mulative dispersion for the male data, the first two canoni- is whether any individuals in a population exhibit both cal vectors account for 75%, and 90% is reached between central and lateral tubercle patches. It is assumed that the fourth and fifth canonical vectors. All individuals in if any individuals in a population have both central and the Río Ucayali and South Central Bolivia samples were lateral patches, all individuals in that population will de- posteriorly correctly identified in the full data classifica- velop both. The data appear to demonstrate a geographic tion matrix, but all individuals from any sample were not break between the Caracas and Northern Guyana samples correctly posteriorly classified in the jackknife classifica- (Table 17). Samples for which all males lack chest tuber- tion matrix. The plot of the first two canonical vectors cles provide no information on whether males develop shows the same level of poor discrimination of the samples number 635 • 23

4 3 2 1 1-A

0 3

Factor 2 -1 1-B -2 -3 -4 2 -5 -5 -4 -3 -2 -1 0 1 2 3 4 5 FIGURE 5. Moderate (3 characters) to high (4 or 5 characters) levels of differentiation between geographically adjacent samples. Factor 1

FIGURE 4. Discriminant function analysis results of first and second canonical vectors for female measurement data of 20 geographic samples (denoted by various symbols). The Northern Magdalena between adjacent geographic samples. The patterns are Drainage (1-­A, triangle with apex at left)/San Andrés/Providencia best visualized by comparing moderate to high (differing (1-­B, triangle with apex at bottom), Gulf of Panama (2, triangle with by three to five characters) levels of differentiation be- apex at top), and Southern Guyana (3, narrow vertical bar) samples tween geographically adjacent samples (Figure 5). Overall, are distinctive from each other. there is considerable differentiation among adjacent sam- ple pairs that are geographically distant from each other. Taking all the differentiation data together (Figure 5), the most significant break between geographically adja- as the female data (results not shown). There are three cent samples is between the Caracas and Northern Guy- samples that are distinct from each other and the rest of ana samples. These two samples differ from each other by the samples: (1) Gulf of Panama, (2) Northern Guyana, four statistically meaningful characters, and the males of and (3) Madre de Dios. the Caracas sample lack lateral patches of chest tubercles The measurement data indicate that the Gulf of Pan- and have two rounded thumb spines whereas the Northern ama sample is the most distinctive as its distinctiveness Guyana sample males have lateral patches of chest spines appears in both the female and male data analyses. and have one slightly chisel-shaped­ thumb spine. This Ca- racas sample break is discrete from the Northern Guyana sample and is most consistent with species level differen- GEOGRAPHIC VARIATION AMONG THE tiation. The Caracas sample character distribution pattern SAMPLES continues along coastal Venezuela to the Estado de Sucre and the Island of Trinidad. There were few locality samples Adult Form Morphological Characters analyzed between southern Anzoategui State in Venezuela and the Guyana and Venezuelan border along the Estado de For the characters analyzed statistically (including dis- Bolívar (Venezuela), most of which lacked sexually active criminant function analyses results), within the three net- adult males. The closest Venezuelan locality sample with works, there is a moderate trend of greater differentiation an adult male with noticeably hypertrophied arms with between geographically distant sample pairs, as would two round thumb spines is from Bolívar State, Santo Tomé be expected. In terms of evaluating geographic variation, de Guayana, San Felix (AMNH 81456) and an adult male the most insightful patterns of differentiation are those with a single slightly chisel-­shaped thumb spine is from 24 • smithsonian contributions to zoology

Bolívar State, 13 km south, 1 km east Puente Cuyuni (KU with L. insularum larvae. The dorsal oral papilla gap of 166408), over 250 km distant from each other. Additional L. insularum is narrower (21–31% of oral disk) than the data are needed from the Venezuelan states of Amazonas gap in L. melanonotus from Costa Rica (40–60%). The and Bolívar to determine the precise distribution of the two L. insularum larval samples from Isla San Andrés (Colom- species involved and whether they occur in sympatry. bia) do not have any noticeable differences with respect to For the northeast set of samples (Southern Costa Rica/ the L. insularum from Costa Rica. Eastern Panama through Caracas) the level of variation Four larval samples are available in addition to the is consistent with intraspecific variation, although there description by Heyer (1970) for this species. All materi- is considerable variation between the Darien and Gulf of als share the following features: sinistral spiracle, median Panama samples (Figure 5C, D). vent, entire oral disk, a single or double row of oral disk In contrast, there is much more variation among the marginal papillae, a tooth row formula of 2/3, and dorsal remaining samples (Figure 5). A clear geographic pat- fin origin at the tail-­body juncture. The larval materials are tern is demonstrated by whether the male thumb spine is from the following localities, referred to henceforth by the strongly chisel shaped and the posterior thigh predomi- specific localities only: USNM 241340: Isla San Andrés, nantly uniform or moderately mottled (Rio Solimões/ San Andrés y Providencia, Colombia; USNM 330406: Amazonas through the Río Ucayali and South Central near Cañas, Guanacaste, Costa Rica; KU 68365: Puerto Bolivia samples) or a weakly chisel-shaped­ thumb spine Cortes, Puntarenas, Costa Rica; and USNM 330837: and a predominantly boldly mottled posterior thigh pat- Rincón de Osa, Puntarenas, Costa Rica. tern (Northern Roraima through Suriname samples). The The two specimens from Isla San Andrés and the morphological data are equivocal in terms of whether the sample from Puerto Cortés are faded. The dorsal bodies break between this clustering of samples represents intra-­ of the Cañas tadpoles are relatively uniform brown; the or interspecific variation. guts are very visible from the lateral view with the por- tion anterior to the guts relatively uniform tan/brown; the Larval Morphology ventral aspect of the body is lighter than the dorsum, but melanophores are present on the portion anterior to the Given the problems of tadpole identification described guts; the tail musculature is uniform brown; the tail fins in Methods and Materials, justification concerning the are much lighter than the musculature but either with scat- identification of the larval samples described below is tered melanophores or a relatively uniform tan pattern. called for. Five species of Leptodactylus occur in Costa The Puerto Cortés tadpole dorsal aspects of the head and Rica: L. insularum, L. fragilis, L. melanonotus, L. savagei, bodies are mottled brown; from the side, the head-bodies­ and L. poecilochilus. All five Costa Rican species have are mottled; in ventral view, the area of the oral disk to the published larval descriptions and figures (L. insularum as guts has scattered melanophores, with the area over the L. bolivianus, L. savagei as L. pentadactylus, L. fragilis guts without melanophores; the tail fins and musculature as L. labialis; Heyer 1970), and the tadpoles of the five are dark, almost uniform, brown. species can readily be identified based on morphological There is variation among larval quantitative character features. Only L. melanonotus larvae might be confused data (Table 18).

TABLE 18. Quantitative larval data. Number in parentheses under Source is number of specimens upon which the data are based. Defi- nitions: ED, eye diameter; HBL, head–body length; ODW, oral disk width; TL, total length; TN, number of A2 teeth in 0.3 mm. A dash (–) indicates the values were not available in the data source.

Locality Source Gosner stages TL, mm HBL/TL, % ED/HBL, % ODW/BL, % TN

Colombia, Isla San Andrés USNM 241340 (2) 37 32.6–35.2 32–34 11–12 27–28 17–19 Costa Rica Heyer, 1970 34 32.0 32–44 5–10 20–28 – Costa Rica, Cañas USNM 330406 (5) 28–31 24.4–28.5 38–42 6–7 23–27 18–22 Costa Rica, Puerto Cortés KU 68365 (5) 32–34 29.3–32.4 37–39 6–7 21–23 18–20 Costa Rica, Rincón de Osa USNM 330837 (5) 25–30 17.4–18.4 44–47 8–10 20–23 25–28 number 635 • 25

The variation of the data described above is difficult due to tape speed variation; the most consistent values are to evaluate in terms of discerning intraspecific from inter- those reported in Table 19. Harmonic structure is present, specific differentiation. At least some of the quantitative but not strongly developed. variation (Table 18) should be correlated with ontogeny as represented by Gosner developmental stages. Head Panama Recording and body lengths were measured differently for the Heyer (1970) data than for the data presented here, which com- Figure 7 promises comparability. If one makes the reasonable as- sumption that all of the Costa Rican data represent a The Barro Colorado Nature Monument recording has single species, those data pretty much encompass all of the calls consisting of single notes. There is sufficient other variation observed among all the larval data. sound in the broadcast channel that the amplitude intensity of the call is not definitively clear; most calls have evidence Advertisement Calls of two primary pulses per note. The first portion of the note (approximately 30%) has a faster rise of frequency than the There are only three relatively noisy recordings on next 50% of the note. The final approximate 20% of the hand for members of the L. bolivianus-­insularum com- note has a noticeable falling frequency. Harmonic structure plex suitable for analysis in addition to the information is present and stronger than in the KU 801 recording. published by Fouquette (1960) (Table 19). Peru Recording Costa Rica Recording Figure 8 Figure 6 Recording USNM 268 cut 11 has calls consisting of 1 Recording KU 801, as indicated previously, has a no- or 2 notes. The quality of the recording is not sufficient to ticeably erratic tape speed, which adds an error factor to evaluate detailed temporal structure. The analyses suggest the quantitative data. The call consists of a single note that the second note for two note calls or the single note composed of 1 or 2 pulses. The recording has two call of one note calls have at least two pulses. The first 20% rates, in the beginning 19 s the call rate is 1.2 calls/s, in the (approximately) of the call has a sharply rising frequency, last 39 s, the call rate is 2.5 calls/s. The first half of the call with a decreased rate of rising frequency for the remainder has a sharper rising frequency modulation than the last of the call. The highest frequency is consistently at the end half of the call. The highest frequency occurs at the end of of the call. There is no visible harmonic structure in the the call or shortly before the end of the call (at about 85% audiospectrogram analyses of the calls; there is too much total call duration). There is considerable variation in noise in the wave form analyses to determine with cer- dominant frequencies among the calls analyzed doubtless tainty whether there is harmonic structure or not.

TABLE 19. Advertisement call data. Definitions: ID, identification; SVL, snout–vent length. A dash (–) indicates values not available from the data source.

Voucher No. Call Beginning Highest Ending Dominant Recording Temperature, SVL, notes/ duration, frequency, frequency, frequency, frequency, Country ID °C mm call Calls/s s Hz Hz Hz Hz Harmonics

Costa Rica KU 801 26 – 1 1.2–2.5 0.08 220–300 1,080–1,200 1,060–1,200 920–1,130 + Panama BCI CD – – 1 1.5 0.10–0.12 110–180 890–1,100 510–600 740–870 + Panama Fouquette, 24 – – 1.9 0.10 200 900 750 – + 1960 Peru USNM 268, 23.5 112 1–2 0.5 0.12–0.16 130–200 730–780 730–780 600–690 ? cut 11 26 • smithsonian contributions to zoology

30 20

e 10 0.001 essur

Pr -10 -20

S 19.466 19.5 19.52 19.54 19.56 19.58 19.6 19.62

5

tz 4

3 Kiloher

2

1

0.000 19.466 19.6 19.7 19.8 19.9 Time in seconds

FIGURE 6. Advertisement call of Leptodactylus, KU recording 801 from just west of Piedras Blancas, Puntarenas, Costa Rica, 8 April 1966, water temperature 78°F.

The available advertisement call data (Table 19) are recovered: subclade A, including samples from Panama, most consistent with two call types representing distinct Venezuela, and Colombia, corresponds to the taxon species: (1) the recordings from Costa Rica and Panama; known as L. insularum; subclade B, found in Peru and and (2) the recording from Peru. Bolivia, represents the taxon known as L. bolivianus; and subclade C is recognized herein as a new species in this complex occurring in the Guiana Shield region. MOLECULAR RELATIONSHIPS The genetic divergences among the four subclades recovered within the L. latrans species complex range A total of 800 aligned base pairs were included in from 9.5 to 14.1% (see Figure 9 for range of divergence the analyses, of them 540 were constant and 260 variable values followed by the mean value below corresponding (172 parsimony informative). Base composition consisted branches). Mean value divergences among the three species of A = 34%, C = 23%, G = 18%, and T = 25%. Thirty-one­ recognized in the L. bolivianus–insularum complex from (3.87%) positions included an indel position in at least other species of the L. latrans species group range from 4.2 one taxon. to 7.2%. Divergences among the three species recognized All analyses recovered a clade that included all rep- in the L. bolivianus–insularum complex are bolivianus– resentative samples of the L. latrans species group (Fig- insularum 4.9%; insularum–sp. nov. 6.2%; and bolivi- ure 9). Within this clade, a subclade consisting of L. latrans anus–sp. nov. 5.1%. In addition, mean divergences among and L. chaquensis is recovered and is the sister group to the sampled species of the latrans species group are insu- the L. insularum-­bolivianus complex clade. Within the larum–latrans = 12.2%; insularum–chaquensis = 12.1%; L. insularum-­bolivianus complex, three subclades are bolivianus–latrans = 10.1%; bolivianus–chaquensis = number 635 • 27

20 e

0.001 essur Pr -20

S 6.74 6.8 6.85 6.9

5

4 tz

3 Kiloher

2

1

0.000 6.74 7 7.2 7.4 7.6 7.8

Time in seconds

FIGURE 7. Advertisement call of Leptodactylus from Barro Colorado Nature Monument CD record- ing, track 46, Panama.

10.8%; latrans–sp. nov. =12.9%; chaquensis–sp. nov. = previously indicated, Marinus Hoogmoed (pers. comm.) 12. 6%; and latrans–chaquensis = 6.8%. The existence of informed WRH at the VI World Congress of Herpetol- an undescribed taxon demonstrated by the genetic diver- ogy in 2008 that he had reexamined frogs from the state sification recovered in this analysis in combination with of Pará, near the state of Amapá and found them to be the subsequent identification of diagnostic morphological members of the Leptodactylus bolivianus complex. These characters supports the recognition of the new species. specimens belong to the new species recognized herein. The genetic diversification among this complex of species Thus the single specimen from Amapá examined in this is consistent with the level of diversification among the study is considered to represent the new species as well. other two species in the L. bolivianus complex as well as There are also a number of localities between the dis- level of diversification between species in the L. latrans tributions of L. bolivianus and the new species for which and those in the L. bolivianus complex. only juveniles and/or females are available and rather arbitrary criteria were used to assign species identifica- tions. The following localities are considered to represent INTERSPECIFIC VARIATION L. bolivianus. COLOMBIA. Vaupes: Río Querari near junction of Río Querari and Río Vaupés. BRAZIL. Ama- Three outlying localities for which there are only ju- zonas: Parque Nacional Pico da Neblina, São Gabriel da venile specimens available were considered problematic Cachoeira; Rio Cuieiras, entering Rio Negro on left bank; as to species identification in the preceding analyses. As Rio Demeni, Barcelos; São Gabriel da Cachoeira; São 28 • smithsonian contributions to zoology

200 e 0.50 essur Pr -200

U S 14.755 14.8 14.85 14.9

5

4 tz

3 Kiloher 2

1

0.000 kHz S 14.755 15 15.2 15.4 15.6 15.8 Time in seconds

FIGURE 8. Advertisement call of Leptodactylus, USNM recording 268 cut 11 from Tambopata, Madre de Dios, Peru.

João, near Tapurucuara. VENEZUELA. Amazonas: Cata- For the male analysis, the first canonical vector ac- rata de Huá (Salto de Huá); Río Casiquiare, 8 miles [12.9 counts for 84% of the cumulative dispersion and the first km] below Orinoco (Canal Casiquiare); Río Mavaca, 108 two canonical vectors account for 100% of the cumulative km SSE Esmeralda. Presumed localities for the new species dispersion. Eighty-­three percent of L. bolivianus, 92% of follow. BRAZIL. Roraima: Uiacás, Rio Urariquera; Serra L. insularum, and 85% of the new species are correctly dos Surucucus; Missão Catrimani; Cachoeira do Cujubim, identified using all data in the posterior classification ma- Rio Catrimani; Parque Nacional Viruá, próximo a Cara- trix. The male data jackknifed classification matrix for caraí; Praia do Guariúba; Santa Maria do Buiaçu. The L. bolivianus, L. insularum, and Leptodactylus, new spe- comparisons that follow include the preceding locality/ cies are 78, 92, and 79% respectively. specimen data, and even if most presumed identifications For the female data discriminant analysis, the cumu- are incorrect, the overall results would not be meaning- lative proportions of total dispersion are identical to the fully impacted. male data (84 and 100%). Seventy-­one percent of L. bo- livianus, 87% of L. insularum, and 80% of Leptodacty- Measurement Data lus, new species are correctly identified using all data in the posterior classification matrix. The female data jackknifed Complete model discriminant function analyses of the classification matrix is 67, 84, and 78% respectively. measurement data for females and males indicate incom- The measurement data indicate that there is differen- plete separation of the three species, with the male data tiation among the three species, but the differentiation is (Figure 10) showing better discrimination than the female not complete enough to be useful as diagnostic informa- data (Figure 11). tion to correctly identify all specimens to species. number 635 • 29

FIGURE 9. Maximum likelihood cladogram of the Leptodactylus latrans species group with other representatives of Leptodactylus species lineages.

4 FIGURE 10. Plot of first two canonical vectors for male measurement data for the three species of the Leptodactylus bolivianus cluster. Dot (●), L. boliv- ianus; ×, L. insularum; +, new species. Centroid confidence ellipses, p = 0.95.

1

OR(2) Morphological Data

CT FA -2 Forty-­four percent of the morphological character comparisons among species pairs are compromised by high observer error (Table 20). The high observer error is accounted for, in part, by the fact that the three species are rather morphologically similar to each other. -5 For the statistically significant character compari- -5 -2 1 4 sons, the most differentiated character states are listed as FACTOR(1) follows. 30 • smithsonian contributions to zoology

7.0 with greater occurrence. Thumb spines, state B: one hy- pertrophied chisel-shaped­ spine; L. bolivianus with exclu- sive occurrence. 4.8 Leptodactylus insularum: New Species Comparisons 2.6

OR(2) Belly pattern B: boldly mottled anteriorly only or with an anterior-posterior­ gradient; new species with greater 0.4 occurrence. Lateral folds, both states equally differenti-

CT FA ated in both species. Chest tubercles state B: central patch of tubercles only; new species with greater occurrence. -1.8 Thumb spines, state A: one modest chisel-shaped­ spine on each thumb; new species with greater occurrence. The male secondary sexual characteristics demonstrate the -4.0 -4.0 -1.8 0.4 2.6 4.8 7.0 greatest diagnostic differentiation among the three species. FACTOR(1) SPECIES ACCOUNTS FIGURE 11. Plot of first two canonical vectors for female measure- ment data for the three species of the Leptodactylus bolivianus clus- There are several characters examined in this study ter. Circle, L. bolivianus; ×, L. insularum; +, new species. Centroid that demonstrate negligible differences among the three confidence ellipses, p = 0.95. species. All three species exhibit the following character states, which are not repeated in the individual species accounts. Dorsal patterns are uniform between the interocular Leptodactylus bolivianus and bar (if present) and the sacrum or with a single chevron L. insularum Comparisons between the interocular bar and sacrum without additional spots lateral to the dorsal chevrons or a complex pattern Dorsal pattern A: uniform between interocular bar with two dorsal chevrons and spotting all over the dorsum, [if present] and sacrum; L. insularum with greater occur- chevrons often elongate and fused with each other and the rence. Lip stripe pattern D: lip region uniform; L. bolivi- interorbital bar. Light subocular patches vary from very anus with greater occurrence. Posterior thigh patterns A distinct to not discernible. The dorsal surface of the shank and B: both species differing by 86% for each state. Belly patterns are uniform or speckled or with short dark trans- pattern C: belly speckled or mottled with same intensity verse bars extending less than halfway across the shank or over entire belly; L. insularum with greater occurrence. with at least one dark transverse bar extending halfway Thumb spines state C: two round spines; L. insularum or more across the shank or with large dark spots in the with 100% occurrence. middle of the upper shank. Dorsolateral folds are distinct and continuous from the supratympanic fold to the end Leptodactylus bolivianus: of the body, or with a brief interruption of the fold, or the New Species Comparisons fold is distinct only from the supratympanic fold to the sacrum, or the fold has several interruptions, or the fold is Posterior thigh pattern A: uniform or variable disper- indiscernible. The toes are fringed. Adult males have either sion of melanophores (weakly to moderately mottled) usu- keratinous tubercles on the tympanum or lack tubercles. ally for entire extent of posterior thigh; L. bolivianus with Specimens from two localities cannot be identified greater occurrence. Belly pattern A: speckled, spotted, or to species: USNJM 162697, 216724–216725, Venezu- mottled anteriorly only or with an anterior-posterior­ gra- ela, Bolívar, Los Patos, 25–28 km SE El Manteco, 350 m, dient; L. bolivianus with greater occurrence. Chin tuber- 7°11N, 62°22W; and UMMZ 195839, Venezuela, Ama- cles: both states equally differentiated in the two species. zonas, San Fernando de Atabapo, 4°3N, 67°28W. The Chest tubercles state B: central patch only; new species specimens involved are metamorphs, juveniles, or females. number 635 • 31

TABLE 20. Statistical comparison of morphological character states for species pairs of the Leptodactylus bolivianus complex; df = degrees of freedom. Significant probability (p) results are in bold italic font.

Leptodactylus bolivianus Leptodactylus bolivianus Leptodactylus insularum Characters and L. insularum and L. new species and L. new species

Dorsal pattern χ2 = 25.186 Character state Character state df = 4 occurrence < observer error occurrence < observer error p = 0.000 Lip stripe χ2 = 33.945 Character state Character state df = 3 occurrence < observer error occurrence < observer error p = 0.000 Subocular spot χ2 = 2.443 χ2 = 1.306 χ2 = 3.711 df = 2 df = 2 df = 2 p > 0.1 p > 0.5 p > 0.1 Tibia pattern Character state Character state Character state occurrence < observer error occurrence < observer error occurrence < observer error Posterior thigh pattern χ2 = 238.297 χ2 = 152.059 Character state df = 2 df = 2 occurrence < observer error p = 0.000 p = 0.000 Belly pattern χ2 = 25.186 χ2 = 76.760 χ2 = 55.100 df = 4 df = 4 df = 4 p = 0.000 p = 0.000 p = 0.000 Dorsolateral folds Character state Character state Character state occurrence = observer error occurrence < observer error occurrence < observer error Lateral folds Character state Character state χ2 = 10.281 occurrence = observer error occurrence < observer error df = 1 p = <0.005 Chin tubercles χ2 = 0.035 χ2 = 8.607 χ2 = 0.489 df = 1 df = 1 df = 1 p > 0.5 p = 0.000 p > 0.5 Chest tubercles χ2 = 0.096 χ2 = 7.747 χ2 = 7.281 df = 1 df = 2 df = 2 p > 0.1 p < 0.025 p < 0.05 Thumb spines χ2 = 18.040 χ2 = 16.333 χ2 = 20.5 df = 1 df = 1 df = 1 p = 0.000 p = 0.000 p = 0.000 Tympanum tubercles Character state Character state Character state occurrence < observer error occurrence < observer error occurrence < observer error

There are additional localities for which the morphologi- Leptodactylus bolivianus Boulenger, 1898 cal data are ambiguous as to which species is involved for Figure 12 samples geographically between the new species and L. in- sularum. The identifications for these localities are based Leptodactylus bolivianus Boulenger, 1898:131. Lectotype: Museo Civico on river drainage distributions, and new data are needed di Storia Naturale de Genova 28875, male, designated by Capocaccia to verify which of the two species occur at the localities (1957:214). Lectotype locality: Barraca (Río Madidi, Bolivia). involved. These latter localities are distinguished on the Leptodactylus romani Melin, 1941:54. Lectotype Göteborgs Naturhistoriska distribution map (see Figure 14) and in the distribution Museum (GNM) 499, juvenile, designated by Heyer (1969:6). Lecto- sections of the following pertinent accounts. type locality: Taracuá, Rio Uaupés, Brazil. 32 • smithsonian contributions to zoology

Etymology. There is no stated etymology in the original description. The species is named for the coun- try of Bolivia from which the specimens were collected. Diagnosis. Most specimens of L. bolivianus have the posterior thigh with a uniform or variable dis- persion of melanophores (weakly to moderately mottled) usually for the entire extent of the posterior thigh; most specimens of L. insularum and L. n. sp. have boldly mottled posterior thighs with large light irregular spots/vermicula- tions on a dark background to the light marks being more extensive than the dark areas, the light markings sometimes coalesced. Leptodactylus bolivianus sexually active adult males have lateral patches of chest tubercles in addition to a central patch; L. insularum only have the central patch of tubercles. The only entirely consistent character diag- FIGURE 12. Photo of Leptodactylus bolivianus from Tambopata, Peru (courtesy of Reginald B. Cocroft). nosing the three species is the adult male thumb spine. The single spine on each thumb in L. bolivianus is extremely chisel shaped (Figure 13B); L. insularum has two rounded, pointed spines per thumb (Figure 13A); L. n. sp. has a sin- gle, modestly chisel-shaped­ spine per thumb (Figure 13C).

AA BB

CC

FIGURE 13. Male thumb spines. A. Leptodactylus insularum, USNM 565028; B. Leptodactylus bolivianus, USNM 280211; C. Leptodactylus new species, USNM 566172. number 635 • 33

Character Summary. The most frequent 15°38S, 64°37W, AMNH 72382–72395; Tumi Chucua, lip stripe pattern consists of a noticeable broad light stripe 170 m, 10°8S, 66°10W, USNM 280211. Cochabamba: extending from the front of the snout to the commissural Puerto Chipiriri, 300 m, 16°32S, 65°16W (coordinates gland, bordered below by continuous or broken markings for mouth of Río Chipiriri), AMNH 72238–72241; Puerto on the upper lip and either bordered above by a dark line San Francisco, 36 km N Chipiriri, 16°30S, 65°20W, under the eye or not. The most frequent posterior thigh KU 135508; Villa Tunari, 10.5 km E, 500 m, 16°59S, pattern is a uniform or variable dispersion of melanophores 65°20W, KU 183010–183011. Pando: Nacebe, 11°0S, (weakly to moderately mottled) usually for the entire ex- 67°25W, NMP6V 72179/3, NMP6V 72179/4. Santa Cruz: tent of the posterior thigh. The most frequent lateral fold Buenavista (= Buena Vista), 400 m, 17°27S, 63°40W, conditions are distinct folds only in the groin region, or UMMZ 66482, 66485, 66535, 83101 (3); Buena Vista, indistinct overall, or indiscernible. The most frequent adult 5 km W, USNM 146513–146514, 146522; Río Chaparé, male chin tubercle condition is presence of the tubercles. 5 km N boca, 15°55S, 64°38W, AMNH 79046–79047; Female (n = 91) SVL 61.2–107.7 mm (m [mean] = Río Ichilo, 34 km S boca Río Chaparé, 16°07S, 64°49W, 85.3), male (n = 25) 79.0–121.5 mm (m = 104.6). Female AMNH 79043; Río Ichilo, ~54 km S boca Río Chaparé, head length/SVL ratio 0.33–0.40 (m = 0.365), male 0.34– 16°10S, 64°40W, AMNH 79048–79051. 0.41 (m = 0.365). Female head width/SVL ratio 0.29–0.37 BRAZIL. Acre: Feijó, 8°9S, 70°21W, MZUSP (m = 0.323), male 0.31–0.39 (m = 0.345). Female eye-­ 106665–106667; Igarapé de Nico, Rio Acre, 9°49S, nostril distance/SVL ratio 0.09–0.12 (m = 0.102), male 67°36W, USNM 202431; Reserva Florestal Humaitá, 0.09–0.11 (m = 0.102). Female tympanum distance/SVL 9°45S, 67°40W, MZUSP 70866; Rio Branco, 60 m, ratio 0.06–0.10 (m = 0.074), male 0.06–0.08 (m = 0.071). 9°58S, 67°48W, MZUSP 6487; Rio Muru, 8°09S, Female thigh length/SVL ratio 0.40–0.53 (m = 0.462), 70°45W, MZUSP 106673–106675; Tarauacá, 190 m, male 0.39–0.51 (m = 0.458). Female shank length/SVL 8°10S, 70°46W, FMNH 83238, MZUSP 64730,106668– ratio 0.44–0.56 (m = 0.516), male 0.46–0.56 (m = 0.502). 106672. Amazonas: Lago do Miuá, Anavilhanas, 3°42S, Female foot length/SVL ratio 0.47–0.60 (m = 0.528), male 62°8W, INPA 1232; [Barcelos, Rio Demeni, 1°0S, 0.48–0.57 (m = 0.526). 62°58W, INPA 9613]; Boca do Acre, 8°45S, 67°24W, Tadpole. Unknown. MZUSP 50230–50248, USNM 202432–202448; Borba, Advertisement Call. Calls of 1–2 notes at a 25 m, 4°25S, 59°35W, MNRJ 3169, 13522, MZUSP rate of 0.5/s; call duration 0.12–0.16 s; call amplitude mod- 56499–56501, USNM 202451–202453; Fortaleza, ulated, each note with loudest portion in middle of note; médio Purus, 6°26S, 64°17W, MZUSP 4474, 4480, call frequency modulated starting at a low frequency (130– 8141; Humaitá, San Miguel, 70 m, 7°31S, 63°2W 200 Hz) followed by a long portion of higher frequency (coordinates for Humaitá), MNRJ 19295–19296; Iga- (730–780 Hz) (Figure 8). Recording USNM 268 cut 11. rapé Belém, Rio Solimões, 3°55S, 69°37W, AMNH Distribution. Central and western portions 97077–97079; [Rio Cuieiras, 2°50S, 60°30W, MZUSP of the Amazon basin in Bolivia, Brazil, Colombia, Peru, 65391, 65394 (metamorphs)]; [São Gabriel da Cachoeira, and Venezuela (Figure 14). Parque Nacional Pico da Neblina, 0°5S, 65°55W, AL Specimens Examined (n = 347). Locali- 2015, INPA 15634, 15637]; [São João, próximo Tapu- ties enclosed by square brackets require validation that the rucuara, 0°33S, 64°57W, MZUSP 37509]; Xerua, lado species is correctly identified. BOLIVIA. Beni: Alejandria, direito, Rio Juruá, 6°8S, 67°50W, INPA 2248. Rondô- Río Mamoré, 12°4S, 65° 9W, AMNH 79053–79066; nia: Abunã, Rio Madeira, 9°42S, 65°23W, AL 122; Beni Biosphere Reserve, El Trapiche/Palm camp, 300 m, Calama, 50 m, 8°4S, 62°52W, USNM 202455; Foz do 14°30S, 66°0W, USNM 306617; Ivon, 11°7S, 66°9W, Jamari, 8°28S, 63°30W, MZUSP 56504–56505; USNM UMMZ 58837; Puerto Almacen (= Mayor Pedro Vaca 202449–202450; Porto Velho, 98 m, 8°46S 63°55W, Diez), 260 m, 14°47S, 64°51W, AMNH 72381; Río INPA 15458, MZUSP 16657; São Carlos, 100 m, 8°28S, Grande, 5 km NW of mouth, 15°51S, 64°39W (coordi- 63°30W, USNM 202454. nates for mouth of Río Grande), AMNH 79042, 79052; COLOMBIA. Amazonas: Leticia, ~100 m, 4°9S, Río Mamoré, Río Ibarre (= Ibare), 14°37S, 64°57W, 69°57W, AMNH 81859–81862, KU 124742, USNM AMNH 79033–79041, 79044–79055; Río Mapiri, 140289, 147029–147030, ICNMHN 46855; Leticia, mouth, below upper Beni, UMMZ 58834 (3), 58835 (2); Parque Nacional Natural Amacayacó, Cabaña Matamata, Rurrenabaque, 220 m, 14°28S, 67°34W, MCZ-­A 10091, 3°58S, 70°10W, IND-­AN 2100; Puerto Nariño, 3°45S, UMMZ 58836 (3), 58838 (4); Santa Rosa, Río Mamoré, 70°22W, KU 153296–153297. Vaupés: [Río Querari near 34 • smithsonian contributions to zoology

FIGURE 14. Distribution of Leptodactylus bolivianus (circles), L. guianensis (squares), and L. insularum (triangles). Question marks indicate localities where additional data are required to confirm the species identities.

junction of Río Querari and Río Vaupés, 1°4N, 69°51W, Madre de Dios: Cocha Cashu, Río Manú between Río USNM 193874 (21.4 mm juvenile)]. ­Panagua and Río Cachira, ~400 m, 11°51S, 71°19W, KU PERU. Locality uncertain, possible Iquitos, AMNH 154880–154885; Cuzco (= Cusco) Amazonico, 200 m, 42890. Huánuco: Tournavista, 1200 m, 8°5556S, 12°35S, 69°5W, KU 194911, 204508, 205143–205168, 74°4219W, MCZ-­A 75025. Loreto: near Contamana, 207732–207734, 209182; Itahuanía, 450 m, 12°47S, Rean Rean, Lago Suhuayo, Río Ucuyali, 7°15S, 74°54W, 71°13W, FMNH 81533; Las Pampas del Río Heath, AMNH 42899, 43416; Domo Santa Clara, Orellana, ~100 160 m, 13°0S, 69°0W, USNM 314902; Pakitza, 350 m, m, 6°54S, 75°4W, USNM 127192–127194; Explorama 11°57S, 71°17W, USNM 342635–342640; Puerto Lodge, junction Río Yanomono and Río Amazonas, Maldonado, 189 m, 12°36S, 69°11W, FMNH 83306; KU 220363; Iquitos, 100 m, 3°46S, 73°15W, AMNH Tambopata Reserve, Explorer’s Inn, 30 km (airline) SSW 42570, MCZ-A­ 100080; Pampas Hermosas (= Pampa Puerto Maldonado, 280 m, 12°50S, 69°17W, USNM Hermosa), Río Cushabatay, 7°12S, 75°18W, AMNH 222277–222280, 247337–247355, 247634–247644, 42141, 42147, 42831, 43120; Roaboya, ~100 m, 7°48S, 268961–268967, 342998, 343243; Tres Chimbadas, ~20 74°52W, AMNH 42603–42604, 42920, 45532; “San min downstream from Explorer’s Inn, 12°47S, 69°17 W, Antonio,” Río Itaya, 3°56S, 73°43W, AMNH 43214. USNM 222275–222276, 222281. Ucayali: Balta, Río number 635 • 35

Curanja, ~300 m, 10°8S, 71°13W, KU 196540–196554; Etymology. The species is named for its dis- Iparía, TNHC 37407; Pucallpa, 150 m, 8°23S, 74°32W, tribution coinciding in large part with the Guiana Shield. FMNH 56306–56307, 56446; Yarinacocha, 100–170 m, Diagnosis. Most specimens of L. guianensis 8°15S, 74°43W, FMNH 45402–45405, 56292, 56295– have boldly mottled posterior thighs with large light ir- 56298, 56301. regular spots/vermiculations on a dark background to the VENEZUELA. Amazonas: [Río Casiquiare, 8 miles light marks being more extensive than the dark areas, the below Orinoco, 3°5N, 65°53W, AMNH 23167]; light markings sometimes coalesced; most specimens of [Río Mavaca, 108 km SSE Esmeralda, 140 m, 2°15N, L. bolivianus have the posterior thigh with a uniform or 65°17W, USNM 216705–216723]; [Río Maturacá, Cata- variable dispersion of melanophores (weakly to moderately rata de Huá (= Salto de Huá), 100 m, 0°48N, 66°21W, mottled) usually for the entire extent of the posterior thigh. USNM 83575]. The only consistent character diagnosing the three species is the adult male thumb spine. The single spine on each Leptodactylus guianensis, new species thumb in L. guianensis is modestly chisel shaped (Figure 13C), the single thumb spine of L. bolivianus is extremely Figure 15 chisel shaped (Figure 13B), and there are two rounded and pointed spines on each thumb in L. insularum (Figure 13A). Holotype. USNM 531509, an adult male Description of Holotype. Figure 16. from Guyana; Rupunini, Iwokrama Forest Reserve, Si- Snout rounded in dorsal and profile views. Canthus rostra- puruni River, Pakatau Camp, 4°4517’N, 59°0128’W, lis indistinct. Loreal concave-­obtuse. Tympanum distinct, 85 m. Collected by Maureen A. Donnelly and Megan greatest diameter about 9/10 eye diameter. Tympanic an- Chen on 15 July 1997. nulus and adjacent area with numerous small keratinized Paratopotyes. USNM 531508 (juvenile fe- tubercles. Vomerine teeth in two strongly arched series, male), 531510–531511 (adult females), same locality and narrowly separated from each other medially, positioned collectors; collected on 15 July 1997, 16 July 1997, and posteriorly to choanae. Each elliptical choana with ante- 21 July 1997 respectively. rior shelf extending over anterior edge of choana. Vocal Paratypes. AMNH 145130 and 145131 slits elongate, parallel to jaw. Vocal sac not visible exter- (adult males, tissue sampled), Guyana: East Berbice– nally. Finger lengths in increasing order V < III < II < IV. Corontyne Region, Dubulay Ranch on the Berbice River, Fingers with lateral fleshy ridges, clusters of small keratin- 60 m, 5°4055N, 57°5132W; collected by Charles J. ized tubercles on upper side of finger II joints and rows Cole and Carol R. Townsend on 30 August 1995. of similar tubercles on the inner ridges of fingers III and All other specimens examined are referred specimens, IV. Finger subarticular tubercles pungent. Inner palmar tu- not paratypes. bercle large, oblong, pungent; outer palmar tubercle large, flat, elliptical, bifid distally, narrowly separated from inner tubercle. Smaller, rather indistinct palmar tubercles proximal to each subarticular tubercle and one indistinct palmar tubercle between metacarpal tubercles. Arms hy- pertrophied. Each thumb with one conical dark keratin- ized spine with a slightly developed chisel-­shaped tip, tip on right thumb spine with three small cusps, tip on left with two small cusps. No large spines on chest. Upper and lower lips with numerous small keratinized tubercles. One central patch of numerous small keratinized tubercles on chest and a pair of tubercle patches on the lateral portions of the chest. Ulnar ridges weakly developed. Dorsal texture smooth anteriorly, with small scattered keratinized tuber- cles on posterior 2/3 of dorsum. A pair of well-­developed dorsolateral folds extending from just posterior to the eye FIGURE 15. Color photo of Leptodactylus guianensis from Co- to the upper leg. A pair of slightly interrupted lateral folds eroeni Island, Suriname (courtesy of Marinus S. Hoogmoed). from above the arm to groin. A pair of elongate glands above arm insertion. Belly disk fold distinct posteriorly 36 • smithsonian contributions to zoology

A A C C

B B

FIGURE 16. Dorsal view of (A) head, lateral view of head (B), and ventral view of the head, hands, and chest (C) of Leptodactylus guianensis, new species, USNM 531509.

only. Scattered keratinized tubercles ventrally in arm in- through nostril to anterior edge of eye. Dark posteriorly sertion area. Toe tips rounded. Toes with well-­developed directed interorbital triangle continuous with chevron in fringes and basal webbing. Subarticular tubercles moder- scapular region, rest of dorsum nondescript. Pair of dark ately pungent. Weak metatarsal ridge. Outer metatarsal oval spots above arm insertion. Posterior portion of lateral tubercle almost rounded, about 1/3 size of prominent, fold distinctly lighter, some lateral warts light highlighted elongately oval inner metatarsal tubercle. Tarsal fold well as well. Chin and anterior chest darkly mottled. Central developed, straight except for most proximal end, extend- chest and anterior belly with light and dark labyrinthine ing about 9/10 length of tarsus. Tarsal fold with scattered pattern, grading posteriorly to almost completely lacking small keratinized tubercles. Tarsal fold abutting inner melanophores except for a few small scattered dots. Poste- metatarsal tubercle. Upper shank surface shagreened with rior thigh mottled, overall dark appearance, with sizeable scattered small keratinized tubercles. Scattered keratinized irregular lighter vermiculations. tubercles in arm insertion area. Outer tarsus and sole of Measurements (mm): SVL 100.5, head length 34.4, foot with scattered small keratinized tubercles. head width 35.5, eye–midnostril distance 10.4, greatest Upper lip with light outlined dark triangular mark, tympanum diameter (including annulus) 7.9, thigh length apex ending just short of eye. Light stripe from below eye 47.2, shank length 50.8, foot length 51.8. becoming very distinct under the tympanum and extending Character Summary. The most frequent to end of commissural gland. Dark stripe from tip of snout lip stripe patterns consist of a noticeable broad light number 635 • 37 stripe extending from the front of the snout to the com- 70724; [Missão Catrimani, 0°27N, 61°41W, MZUSP missural gland, bordered below by continuous or broken 68734–68736]; [Parque Nacional Viruá, Roraima, próx- markings on the upper lip and either bordered above by imo a Caracaraí, Igarapé Viruá, 1°0N, 61°15W, AL a dark line under the eye or not; or lip region uniform 12241–12242]; [Praia do Guariúba, 0°20N, 61°44W, (about same intensity as background dorsal color) with MZUSP 67237–67240, USNM 302471–302473]; [Santa or without a dark stripe under the eye and/or dark bars/ Maria da Boiaçu, MZUSP 67330, 67332–67335, 68296]; mottling on the upper lip. The most frequent posterior [Serra do Surucucus, 2°47N, 63°40W, MZUSP 65817]; thigh pattern is boldly mottled with large light irregular [Uiaca, Rio Uraricoera (= Uaicás, Rio Urariquera), 3°33N, spots/vermiculations­ on a dark background to the light 63°11W, AMNH 100009–100010]. marks being more extensive than the dark areas, the light GUYANA. Region unknown: Anowini Creek, Es- markings sometimes coalesced. The most frequent lateral sequibo River, UMMZ 79474; Essequibo River, AMNH fold conditions are distinct folds only in the groin region, 49257. Barima–Waini Region: Maburama, 100 m, or indistinct overall, or indiscernible. The most frequent 8°12N, 59°47W, UTA 55504. Cuyuni–Mazaruni Re- adult male chin tubercle condition is absence of tubercles. gion: Blackwater Creek, 6°5N, 59°59W, AMNH Female (n = 52) SVL 66.0–109.2 mm (m [mean] = 43670; Bartica, 3 m, 6°24N, 58°37W, MCZ-­A 50704; 88.2), male (n = 34) 79.5–109.5 mm (m = 94.8). Female Kalacoon, Mayarum River (= Mazaruni River), AMNH head length/SVL ratio 0.33–0.40 (m = 0.359), male 0.33– 3990; Kartabo, 100 m, 6°21N, 58°41W, AMNH 10397, 0.39 (m = 0.365). Female head width/SVL ratio 0.31–0.37 10423, 11709; Paruima (= Paruima Mission, Kama- (m = 0.327), male 0.32–0.39 (m = 0.352). Female eye– rang River), 614 m, 5°48N, 61°1W, UMMZ 85151, nostril distance/SVL ratio 0.09–0.11 (m = 0.099), male 58152(2). Demerara–Mahaica Region: Georgetown, sea 0.09–0.12 (m = 0.101). Female tympanum diameter/SVL level, 6°48N, 58°10W, AMNH 5064. East Berbice– ratio 0.06–0.08 (m = 0.069), male 0.06–0.08 (m = 0.072). Corontyne Region: Dubulay Ranch on the Berbice River, Female thigh length/SVL 0.42–0.52 (m = 0.476), male 60 m, 5°4055N, 57°5132W, AMNH 145130–145131, 0.44–0.53 (m = 0.484). Female shank length/SVL ratio 145133–145134. Potaro–Siparuni Region: Magdalen’s 0.48–0.57 (m = 0.529), male 0.50–0.56 (m = 0.525). Fe- Creek Camp, near (~300 yards) [275 m] NW bank of male foot length/SVL ratio 0.48–0.58 (m = 0.527), male Konawaruk River (~25 miles [40 km] linear) WSW Ma- 0.46–0.57 (m = 0.527). bura Hill, 120 m, 5°137N, 59°243W, AMNH 166359– Tadpole. Unknown. 166366. Upper Demerara–Berbice Region: Berbice River Advertisement Call. Unknown. camp at ~18 miles [30 km] (linear) SW Kwakwani (~2 Distribution. Lowland portions of Guiana miles [3.2 km] downriver from Kurundini River conflu- Shield regions of Guyana, Suriname, Venezuela, and adja- ence), 60 m, 5°56N, 58°1414W, AMNH 166341– cent Brazil (Figure 14). 166349, 166356–166358; Malali (= Malali rapids), 100 Specimens Examined (n = 226). Locali- m, 5°37N, 58°22W, AMNH 45773; Penal Settlement, ties enclosed by brackets require validation that the spe- 6°24N, 58°39W, AMNH 10422; Rockstone, 100 m, cies is correctly identified. BRAZIL. Roraima: Boa Vista, 5°59N, 58°33W, FMNH 26633–26636, UMMZ 80442; 90 m, 2°49N, 60°40W, MZUSP 62482, 65508–65510, Wismar, ~100 m, 6°0N, 58°18W, UMMZ 76681(2). 66053; BR-­174 at km 110 (Rio Surumu), 3°25N, Upper Takutu–Upper Essequibo Region: North of Acarahy 60°18W, MZUSP 60616; Marco da Fronteira BV-­8, Mountains, west of New River, 1°50N, 57°30W, KU 4°28N, 61°8W, INPA 1306, MZUSP 65760–65761, 69667–69686; Iwokrama Forest Preserve, Burro-Burro­ 65783–65786, USNM 302445; [Cachoeira do Cuju- river, Burro-Burro­ Camp, 83 m, 4°4352N, 58°512W, bim, Rio Catrimani, 1°45N, 62°17W, MZUSP 58533]; AMNH 164217–164218, USNM 531502; Iwokrama For- ­Cachoeira do Paredão, 2°57N, 61°35W, MZUSP 70634; est Preserve, Kabocalli Camp, 101 m, AMNH 164219– Colonia Apiaú, 150 m, 2°38N, 61°12W, MZUSP 65923, 164221; Iwokrama Forest Preserve, ~3 miles [4.8 km] 65948, 65953, 66317–66319, 68632–68633, USNM S Kurupukari Base Camp on Georgetown-­Lethem road, 302268–302270; Fazenda Salvamento, 3°20N, 61°24W, Three Mile Camp, 102 m, 4°3759N, 58°4252W, MZUSP 66031; Igarapé Cocal, near Tepequém, 3°45N, USNM 531501; Iwokrama Forest Preserve, 5 hours 61°44W, MZUSP 66008, 67004; Ilha do Maracá, 3°25N, (downstream) Kurupukari Base Camp on Essequibo 61°40W, MZUSP 60628, 62420–62426, 65548–65558, River, Kabocali Camp, 101 m, 4°176N, 58°3034W, 65573, 65613, 65670–65672, 65698–65700, 65703– USNM 531503–531507; Iwokrama Forest Preserve, Sipu- 65705, 65735–65736, 65743–65744, 70649–70650, runi River, Pakatau Camp, 85 m, 4°4517N, 59°128W, 38 • smithsonian contributions to zoology

USNM 531508 (Paratype, juvenile female), 531509 (Holotype, male), 531510–531511 (Paratypes, females); Iwokrama Forest Preserve, Pakatau Creek, 85 m, 4°45N, 59°1W, AMNH 164223–164227; northern Rupununi Sa- vanna, Karanambo (on Rupununi River), McTurk Ranch, 110 m, AMNH 136033–136034; Karanambo Ranch, Maricuba Lake, ~100 m, 3°458N, 59°1836W, USNM 497730; Kuyuwini, 1°55N, 59°15W, AMNH 43439, 43684, 46278–46279, 46286; Moco-Moco­ (= Moco- moco), 29 km SE Lethem, base of Kanuku Mountains, 100 m, 3°1812N, 59°390W, USNM 497731; Mocho Mocho (= Moco Moco, = Mocomoco), USNM 146369– 146374; Onora River (= Unorowo River), tributary of Es- sequibo River, 1°35N, 58°30W, AMNH 49331–49335, 53440–53445. FIGURE 17. Photo of Leptodactylus insularum from San Andrés, SURINAME. Brokopondo: Brownsberg Nature Park, Colombia (courtesy of Roy W. McDiarmid). near Mazaroni top, ~450 m, 4°55N, 55°12W, AMNH 87757; Loksietattie, Saramacca River, 5°2N, 55°32W, FMNH 134739. Commewijne: Plantation Ma Retraite, sea level, 5°52N, 55°8W, AMNH 125820. Nickerie: data. Subsequently, 11 of the 12 original syntypes were Kaiserberg airstrip, Zuid River, FMNH 128810–128818, exchanged or had new MCZ catalogue numbers assigned 128820–128825, 128919, 128926, 128929, 128944. to them. There remains one specimen with MCZ number Para: near Zanderij, sea level, 5°27N, 55°12W (coordi- 2424. As the specimen bearing number 2424 is the only nates for Zanderij), USNM 159513–159514. Saramacca: specimen currently with that number in the MCZ collec- Foengoe Island Airstrip, Raleigh cataracts, Coppename tion, we hereby designate MCZ 2424 as the lectotype of River, 50 m, 4°44N, 56°12W, AMNH 87758. Sipaliwini: Leptodactylus insularum Barbour, 1906. Drietabbetje, Tapahony River, 4°7N, 54°40W, MCZ-­A Lectotype Description. Adult female 97282; Marapi LLB Station on Covantyn River, 5°N, (Figure 18) with pigmented ova. Snout nearly rounded 57°W, MCZ-A­ 110622. Wanica: Pad (Path) van Wanica, in dorsal view, rounded in profile view. Canthus rostra- 5°47N, 55°20W, KU 204508. lis indistinct. Loreal concave-­obtuse. Tympanum distinct, VENEZUELA. Bolívar: Arabupo (= Arabopo), greatest diameter about 7/10 eye diameter. Tympanic 1,200–1,300 m, 5°6N, 60°44W, UMMZ 85198; ~35 km annulus smooth. Vomerine teeth in arched series, apices S El Manteco, 350 m, 5°6N, 62°37S, TCWC 60158; 13 anterior, narrowly separated from each other medially, km S, 1 km E Puente Cuyuni, 140 m, 6°43N, 61°35W posterior to choanae. Choanae damaged, shape uncertain. (coordinates for Puente Cuyuni), KU 166405–166408; Finger lengths II = IV > III ≈ V. Fingers with weak lateral Santa Lucia de Surukun, 45 km NE Icabaru, 851 m, ridges or smooth. Finger subarticular tubercles rounded, 4°33N, 61°25W, USNM 216726. not pungent. Inner palmar tubercle large, ovate, promi- nent; outer palmar tubercle present, somewhat indistinct, Leptodactylus insularum Barbour, 1906 right hand tubercle weakly heart-­shaped, left present, shape unclear. Dorsum smooth. A pair of distinct dark Figure 17 dorsal folds from just behind eye to sacrum, folds indis- tinct from sacrum to juncture of leg and body. Supratym- Leptodactylus insularum Barbour, 1906:228. Type locality: Saboga Island, panic fold dark brown from eye to behind jaw. Short dark Panama. lateral folds extending from supratympanic fold at level of posterior tympanum to ending shortly after posterior Barbour (1906) described Leptodactylus insularum sp. arm insertion. No distinct glands. Belly disk fold indis- nov. based on a lot of 12 specimens with the single MCZ tinguishable. Arms and body lacking tubercles. Toe tips number 2424 from Saboga Island in the Gulf of Panama. round, swollen, just larger than toes immediately behind The type description is not based on a single specimen tips. Toes with lateral fringes and weak basal webbing be- but is a composite description including male and female tween toes I–IV. Toe subarticular tubercles large, rounded. number 635 • 39

FIGURE 18. Dorsal and ventral views of the lectotype of Leptodactylus insularum, MCZ 2424.

Weak outer metatarsal ridge continuous with outer ridge Measurements (mm): SVL 83.8, head length 29.2, on toe V. Outer round metatarsal tubercle about ¼ size of head width 29.1, eye–nostril distance 7.8, tympanum di- oval inner metatarsal tubercle. Distinct tarsal fold lack- ameter (including annulus) 5.6, thigh length 38.8, shank ing tubercles, extending from and continuous with inner length 42.3, foot length 24.6. metatarsal tubercle to about 9/10 length of tarsus. Upper Etymology. Barbour did not specifically in- thigh surfaces smooth. Upper shank surfaces with numer- dicate the derivation of the species name. All of the speci- ous small white tubercles. Outer surface of tarsus with mens indicated in the original description are from islands small white tubercles, a few of which with brown keratin- in the Gulf of Panama, hence the specific name. ized tips. Sole of foot smooth. Diagnosis. Most specimens of L. insularum Face with indistinct markings, a slightly darker indis- have boldly mottled posterior thighs with large light ir- tinct interrupted stripe from nostril to eye, postocular area regular spots/vermiculations on a dark background to darker brown than upper lip area. Dorsum brown with the light marks being more extensive than the dark areas, indistinct darker marks scattered between dorsolateral the light markings sometimes coalesced; most specimens folds. Flanks with irregular dark brown spots on tan to of L. bolivianus have the posterior thigh with a uniform brown background. Belly with small tan dots visible under or variable dispersion of melanophores (weakly to mod- magnification, overall belly very light tan, throat slightly erately mottled) usually for the entire extent of the thigh. darker. Posterior thigh surfaces with a pattern of large Leptodactylus insularum adult males only have central darker and lighter brown mottling. chest patches of tubercles; L. bolivianus males have lateral 40 • smithsonian contributions to zoology

FIGURE 19. Illustration of lateral view of Leptodactylus insularum tadpole, USNM 576264, Costa Rica, Guanacaste, ~12 km (airline) SSW Cañas, scale bar = 5 mm.

patches of chest tubercles in addition to the central patch. Tadpole. Figures 19, 20. Exotrophic, lentic, The one entirely consistent character diagnosing the three benthic guild member (McDiarmid and Altig, 1999, guild species is the adult male thumb spine. The two spines on IIA1); oral disk anteroventrally positioned, entire (not each thumb of L. insularum are rounded and pointed (Fig- emarginated), moderate anterior papilla gap (~31% oral ure 13A); the single spine on each thumb of L. bolivianus disc width), single and/or double row of marginal papil- and L. guianensis is modestly to extensively chisel shaped lae; tooth row formula 2/3; spiracle sinistral; vent median; (Figure 13B, C). dorsal fin origin at tail-­body junction; dorsum with a rel- Character Summary. The most frequent atively uniform profusion of melanophores, venter with lip stripe pattern consists of lip region uniform (about light to moderate profusion of melanophores, venter with same intensity as background dorsal color) with or with- light to moderate profusion of melanophores from oral out a dark stripe under the eye and/or dark bars/mot- disk to guts, scattered melanophores over guts, tail muscu- tling on the upper lip. The most frequent posterior thigh lature uniform moderate to dark brown, tail fins ranging pattern is boldly mottled with large light irregular spots/ vermiculations on a dark background to the light marks being more extensive than the dark areas, the light mark- ings sometimes coalesced. The most frequent lateral fold condition is fold distinct or slightly interrupted from su- pratympanic fold to leg. The most frequent adult male chin tubercle condition is presence of tubercles. Female (n = 224) SVL 60.4–99.1 mm (m [mean] = 81.3), male (n = 216) 66.0–104.6 mm (m = 86.8). Female head length/SVL ratio 0.33–0.43 (m = 0.365), male 0.32– 0.41 (m = 0.364). Female head width/SVL ratio 0.30–0.38 (m = 0.334), male 0.32–0.40 (m = 0.353). Female eye– nostril distance/SVL ratio 0.08–0.12 (m = 0.102), male 0.09–0.12 (m = 0.102). Female tympanum diameter/SVL ratio 0.05–0.08 (m = 0.070), male 0.06–0.08 (m = 0.070). Female thigh length/SVL ratio 0.36–0.58 (m = 0.452), male 0.39–0.51 (m = 0.451). Female shank length/SVL FIGURE 20. Illustration of oral disk of Leptodactylus insularum ratio 0.44–0.58 (m = 0.495), male 0.43–0.58 (m = 0.489). tadpole, USNM 576264, Costa Rica, Guanacaste, ~12 km (airline) Female foot length/SVL ratio 0.44–0.57 (m = 0.506), male SSW Cañas, scale bar = 1 mm. 0.42–0.59 (m = 0.503). number 635 • 41 from same pattern as tail musculature to ventral fin with 2739, 2742; Puerto Boyacá-­Puerto Romero carretera, km a gradient of no melanophores next to body to uniform 29–35, 300–320 m, ICNMHN 38438–38440; Puerto brown around midfin; maximum total length stage 37, Boyacá, Insp. Puerto Romero, ­ICNMHN 44726–44730; 35.2 mm (based on voucher specimens KU 68365, USNM Puerto Boyacá, Vereda El Okal, Quebrada La Fiebie, km 241340, 330406, 330837, 576264). 29–30 Otanche road, 280 m, ICNMHN 45427–45437, This description overall agrees with the previous one 45433, 45438. Caldas: Finca Tintiná, 5°1429.5N, reported by Heyer (1970). 75°4111.1W, MPUJ 1827; Guarinocito, 280 m, 5°23N, Advertisement Call. Calls of single notes 74°43W, AMNH 81855–81858, CMB 548–549, MPUJ at rates of 1.2–2.5/s; call duration 0.08–0.12 s; call am- 374; Reserva Riomanso, 5°40N, 74°46W, MPUJ 3016; plitude modulated, loudest portion from about midcall to Sitio La Turqueza, 5°19.9N, 75°425.5W, MPUJ 4/5 call duration; call frequency modulated beginning at 1825–1826, 2075; 15.6 km carretera de la Victoria a la 110–220 Hz with the highest frequency starting about at Central, ICNMHN 34607–34608. Casanare: Orocué, 1/3 call length of 890–1,200 Hz (Figures 6, 7) (based on 200 m, 4°48N, 71°20W, ICNMHN­ 34683; San Luis voucher recordings from Fouquette, 1960; Ibáñez et al., de Palenque, 5°25N, 71°40W, ICNMHN 2664. César: 1999; and KU recording 801). Finca El Diamante, 15 km south of Bosconia, 9°58N, Distribution. Mainland Pacific versant of 73°51W (coordinates for Bosconia), USNM 200350– Costa Rica, throughout lowland Panama, Caribbean drain- 200353; Casa­cara, Municipio Codazzi, 9°20N, 74°45W, ages of Colombia and Venezuela, Trinidad (Figure 14). ICNMHN 39935–39937; Finca Mundo Nuevo, 9°45N, Specimens Examined (n = 1,018). CO- 73°33W, MPUJ 1388–1390; Valle Dupar (= Valledu- LOMBIA. Antioquia: Chigorodó, ~100 m, 7°41N, par), Santa Marta mountains, 202 m, 10°29N, 73°15W, 76°42W, AMNH 76186, USNM 151887; El Cinco, UMMZ 54597–54598. Chocó: Acandí, corregimento Gil- 7°4535.4”N, 74°4649.1W, MPUJ 2073–2074; Jericó, gal, 2000 m, ICNMHN 18112–18114; Parque Nacional 1967 m, 5°47N, 75°47W, MCZ-A­ 24886; Medellin, Natural ­Katios, 7°50N, 77°7W, ­ICNMHN 3224, 47167– 1500 m, 6°15N, 75°35W, AMNH 38828; San Luis, cor- 47174, 47177–47186, IND-AN­ 2000–2002; Río Sucio regimiento La Danta, near Puerto Carreño, ICNMHN Hacienda Sautatá, 20 m, 7°500N, 77°45W, ICNMHN 17225; Nechí, 100 m, 8°7N, 74°46W, FMNH 54568, 187–192; Río Sucio, Vereda Las Teresitas, 7°925N, 54571; 20 km abajo del [below the] Río Nechí, frente 77°2815W, ICNMHN 47188. Córdoba: Ayapel, casco al caserio [in front of the village] Bijagual, 7°5013N, Urbano, 8°1858N, 75°830W, ICNMHN 48165; 74°4743.7W, MPUJ 2840; Urabá, Río Currulao, 50 m, Ayapel, corregimiento Nariño, caserío Playa Blanca, 7°0N, 75°30W, FMNH 63847. Arauca: Estación Cravo 8°1747N, 74°5846W, ICNMHN 48166; Ayapel, Ha- Norte region, ICNMHN 26833–26841, 27226–27231, cienda La Balsa, 8°185N, 75°224W, ICNMHN­ 48167; 27680–27684, 27890. Atlántico: Barranquilla, Airport Ayapel, Hacienda Quibrache, 8°1747N, 74°5846W, Soledad, sea level, 10°55N, 74°46W, USNM 152641; ICNMHN 48163; Lórica, Bosque Ita, 20 m, 9°16N, Municipio Los Pendales, 10°37N, 75°13W, ICNMHN 75°49W, ICNMHN 48172; Lórica, Estación Piscicola, 42275; Estación Piscicultura de Soplaviento, 10°2336N, CVS, 20 m, 9°13N, 75°50W, ICNMHN 48168–48170, 75°828W, IND-­AN 567, 569, 572, 576, 579–581; 48320; Lórica, La Montaña del Mono, 20 m, 9°13N, Puerto Colombia, 10°5928N, 74°5743W, ICNMHN 75°50W, ICNMHN 48171; Monte Libano (= Mon- 1020; Santa Lucia, 10°21N, 74°58W, ICNMHN 46136; telíbano), 8°5N, 75°29W, ICNMHN 42755–42756, Tubará, corregimiento Cuatrobocas, 10°55N, 73°57W, MZUSP 106090–106091; Montería, Hacienda El Dilu- ICNMHN 49010. Bolívar: no other locality data, MPUJ vio, 64 m, 8°4318N, 75°5926W, ICNMHN 48159– 2187–2190; Isla Fuerte, 9°25N, 76°10W, FMNH 48161; Municipio de Montería, ICNMHN 19550–19551; 152371; San Cristóbal, Canal del Dique, ~100 m, 9°53N, Montería, Maracayo, 8°2432, 75°5326W, ICNMHN 75°15W (coordinates for San Cristóbal), AMNH 75742– 48157–48158; Montería, Quebrada Betancí, 8°2244N, 75746, 75748; Hacienda El Ceibal, 80 m, 10°37N, 75°4955W, ICNMHN 48156; Pueblonuevo, Hacienda 75°14W, ICNMHN 44117–44135; Hacienda La Toronto, 8°24N, 75°17W, ICNMHN 48173; Represa de Aguada, 22 km from San Onofre, USNM 152653; Santa Urrá, 70–150 m, 8°14N, 76°1253W, ICNMHN 41313; Rosa, 600 m, 7°58N, 74°3W, ICNMHN 2102, 2105– Río Manso, tributary upper Río Sinu, 7°41N, 76°10W, 2106, 2108, 2245–2246, 2262–2287, 2289. Boyacá: Las USNM 151031–151033; Tierra Alta, 70 m, 8°11N, Mercedes, Hacienda Los Balcones, 380 m, ICNMHN­ 76°4W, FMNH 61807–61812, KU 144967–144977, 38433–38437; Muzo, 1240 m, 5°33N, 74°6W, IND-­AN UMMZ 132466(4). Cundinamarca: Beltrán, 230 m, 42 • smithsonian contributions to zoology

4°47N, 74°47W, ICNMHN 2339–2343, 2603, USNM San Andrés y Providencia: Isla de Providencia, 13°21N, 145708–145742, 147254; Girardot, ~300 m, 4°18N, 81°22W, MCZ-A­ 19261–19266, USNM 76956–76957, 74°48W, AMNH 71575–71577, 75158, USNM 146216– 146214–146215; Isla de Providencia, Agua Mansa, 146218, 146393, 146467; Girardot, Isla El Sol, al Este ­IND-­AN 1096–1102; Isla de Providencia, reservoir, 80 m, de la confluencia del Río Bogotá, 310–320 m, ICNMHN ICNMHN 33421–33427; Isla de San Andrés, 12°32N, 273–275, 277–279, 281–293. Guajira: Municipio Patom- 81°42W, AMNH 71578–71580, KU 265811–265812, ino, 11°148.1N, 73°337.3W, MPUJ 2897; Nazareth MCZ-­A 17370–17375; Isla San Andrés, near Punta (= Nazaret), 200 m, 12°11N, 71°17W, USNM 115112– Paraiso, UMMZ 127882. Santander: El Centro, 100– 115114, 194735; near Piojo, USNM 152676–152677; 200 m, 6°57N, 73°46W, FMNH 81759, 81761, USNM Río Barbacoas, Arroyo de Arenas, 11°25N, 73°5W, 144832–144837. Sucre: Hacienda La Estanzuela, 4 km UMMZ 54601; Riohacha, corregimiento Calabacito, east of Tolú, 9°31N, 75°32W, USNM 200343–200347; 130 m, 11°33N, 72°55W (coordinates for Riohacha), Municipio San Juan de Onofre, corregimiento de Berru- ICNMHN 11859–11865. Huila: Campoalegre, 530 m, gas, Finca Los Morros, 5 m, ICNMHN 43006; Sincelejo ICNMHN 9365; Finca La Tribuna, 3°44N, 75°223W, to Tolú, USNM 150743–150749. Tolima: El Cardonal, MPUJ 3573–3576; Finca San Rafael, 3°315N, 5°513N, 74°4153W, MPUJ 1402, 1405, 1409, 1416, 75°625.5W, MPUJ 1786–1787; Parque Nacional Natu- 1418–1419, 1421; Espinal, Magdalena Valley, 200 m, ral Cueva de los Guacharos, Municipio Acevedo, 1200 m, 4°7N, 74°53W, MCZ-­A 15067, 15071; Finca El An- 1°35N, 76°1W, USNM 148859. Magdalena: Balneario tonio, 3°4737.8N, 75°187W, MPUJ 1788; Guayabal, El Rodadero de Guaira, 11°12N, 74°13W, KU 169075– 4°58N, 74°54W, MPUJ 1391–1394, 1398; Honda, 169089; Bolívar, Santa Marta mountains, 10°21N, 230 m, 5°12N, 74°45W, AMNH 22604, UMMZ 76073; 74°9W, UMMZ 54596; Ciénaga, sea level, 11°1N, 4 km northwest of Honda, 5°14N, 74°46W, AMNH 74°15W, USNM 144157–144578; Don Diego, sea level, 57549–75751; Municipio Margarita, ICNMHN 42765– 11°15N, 73°42W, UMMZ 48118, 48119(3), 48488(10); 42766; Melgar, orillas Río Sumapáz, 400 m, 4°12N, Fundación, 60 m, 10°26N, 74°7W, MCZ-­A 8966–8967, 74°39W, ICNMHN 1422, 18143; Municipio Venadillo, UMMZ 48106–48108, 48110–48111, 48113–48117; 4°4317N, 74°5554W, ICNMHN 43172. Isla Salamanca–Los Cocos, sea level, 10°59N, 74°27W, COSTA RICA. Guanacaste: Estación Experimen- IND-­AN 622–626, 6629; Mamatoco (= Manatoca, Ma- tal Enrique Jiménez Nunez, 13.6 miles SW Cañas, 20 m, matoca), 11°15N, 74°9W, UMMZ 48120; Minca, on 10°20N, 85°9W, USNM 219762–219763; Río Bebe- road to San Lorenzo, 11°12N, 74°4W, USNM 150926; dero, 5 km S Bebedero, 5 m, KU 65653–65656. Pun- Parque Nacional Natural Tayrona (= Tairona, Tayruna), tarenas: Dominical, ~6.4 km N, 9°17N, 83°51W, KU IND-­AN 2462, 2523; Parque Nacional Natural Tayrona, 204367; Golfito, 10 m, 8°38N, 83°11W, KU 32294– Arrecifes, 11°1928.5N, 73°5743.7W, IND-AN­ 2483; 32304, 65651–65652, 86292; Palmar (= Palmar Sur), Parque Nacional Natural Tayrona, Cañaveral, sea level, 8°58N, 83°27W, KU 32293, 65648; Palmar Sur, 5.6 km 11°19N, 73°56W, IND-AN­ 2441–2442, 2488–2490; SE, 8°56N, 83°25W, KU 65646; Palmar Sur, 7 km SE, Parque Nacional Natural Tayrona, 2–3 km cabaña Ca- 8°56N, 83°25W, KU 65647; Parrita, 9°30N, 84°19W, ñaverales, ICNMHN 13611–13613, 13659; Parque MZUSP 106680; Rincón de Osa, 8°42N, 83°29W, KU Nacional Natural Tayrona, Cinto, 11°20N, 74°4W, 102098, USNM 227621–227624; San Isidro del General, IND-­AN 2443, 2446; Parque Nacional Natural Tayrona, 15–20 miles [24–32 km] W on Dominical road, 9°16N, trail between Cañaveral and El Pueblito, USNM 200348– 83°50W, KU 32392; Villa Neilly, 1.6 km WNW, 75 m, 200349; Parque Nacional Natural Tayrona, El Cedro, 8°40N, 82°55W, KU 65649–65650; Volcan Buenos 11°1954.67N, 74°054.31W, IND-AN­ 2463; Parque Aires, 1 mile [1.6 km] E, Cone Finca, 9°13N, 83°25W, Nacional Natural Tayrona, within 2 km of El Cedro sta- UMMZ 117577. tion, USNM 200354; Parque Nacional Natural Tayrona, PANAMA. Province unknown: Between Campana and near La Gayraca, USNM 200356; Parque Nacional Natu- La Venta, USNM 103481; Canal Zone, KU 67947; Istmo ral Tayrona, Los Naranjos, 150 m, 11°18N, 73°54W, do Panamá, MZUSP 5454; Río Morte Arnade,USNM IND-­AN 2482; Río Buritaca, 11°16N, 73°47W (co- 53928; Tabernilla, FMNH 16683. Chiriquí: David, 0.7 ordinates for mouth of river at Atlantic Ocean), USNM miles [1.1 km] north, 0.2 miles [0.3 km] west, 8°26N, 150992–150994; Río Frio, 30 m, 10°55N, 74°10W, 82°26W, AMNH 69714; Progreso, 8°27N, 82°50W, MCZ-­A 16067–16068; between Santa Marta and Mam- UMMZ 58153–58165; Puerto Armuelles, 8°17N, atoco (Manatoca, Mamatoca), USNM 150893–150924. 82°52W, MCZ-A­ 24236–24238; Puerto Armuelles, 7.5 number 635 • 43

km N, 8°19N, 82°49W, KU 107163–107197. Coclé: 41711; Yavisa, across stream from, 8°11N, 77°41W, Penonomé, 6 km SSW, 30–70 m, 8°31N, 80°22W, KU AMNH 41145. Herrera: Parita, 8°0N, 80°31W, USNM 107203; Río Chorrera, 8°18N, 80°23W, USNM 53885; 127262–127286; 5 km northwest Santa María, near Río Valle de Anton, 560 m, 8°37N, 80°8W, USNM 139703. Santa María, 8°8N, 80°41W, KU 107202. Los Santos: 6 Colón: Achiote, 9°1322N, 80°19W, KU 76514–76516; km (road) southwest Los Santos, 7°56N, 80°25W (co- France Field, 9°22N, 79°52W, MCZ-A­ 10143–10144; ordinates for Los Santos), USNM 203649; Tonosí, 40 m, Frijoles, lake at, 9°10N, 79°50W, FMNH 6003–6004; 7°24N, 80°27W, KU 108616. Panamá: Ancon, 8°58N, Gatun, 9°16N, 79°55W, FMNH 16658. Darién: Camp 79°33W, MCZ-­A 10142; Balboa, 8°57N, 79°34W, KU Townsend, Camp Creek, 8°10N, 77°42W, AMNH 107204; Barro Colorado Island, 9°11N, 79°57W, FMNH 40931, 40998–40999, 41027–41031, 41033–41037, 173206, MCZ-­A 19207; Briya Point near Palo Seco, near 41063, 41727; Camp Townsend, below Yavisa, Río Chu- Balboa, 8°55N, 79°34W (coordinates for Palo Seco), cunaque on Río Darién, 8°10N, 77°42W, AMNH 41715; MCZ-­A 10145; Bruja Point, MCZ-­A 10669, 11750; Co- Cana, 460 m, 7°4555N, 77°416W, USNM 50222; Can- coli, 8°59N, 79°35W, USNM 143347; Fort Clayton, clones, 8°20N, 77°46W, UMMZ 132664, 132667(5), 9°0N, 79°45W, MCZ-­A 10141; Fort Kobbe, 8°55N, 132668–132669; Canglon, camp site below, Río Chu- 79°35W, USNM 193348; Howard Air Force Base, cunaque, AMNH 41128, 41130–41131; Chucunaque, 8°55N, 79°36W, USNM 193337, 193339; Isla Chapera, Río Chucunaque, AMNH 41121–41122; Camp Chu- 10 m, 8°35N, 79°2W, AMNH 87337–87339; Isla Can- cunaque, ~0.5 miles [0.8 km] S, AMNH 40614–40615, tadora, 8°38N, 70°02W, AMNH 87334–87336; Isla Sa- 40643–40644, 40699, 40713–40715, 40929, 41088– boga, 10 m, 8°37N, 79°4W, AMNH 87340–87344; Juan 41089, 41734, 41766, FMNH 170431, KU 80380–80384, Mina, 25–30 m, 9°10N, 79°39W, FMNH 57537–57539, 107209–107222, UMMZ 125006, 137749–137750, 60488, 67883, KU 107206, UMMZ 135375; Los Overos, USNM 140575–140583; Jaqué, 7°31N, 78°10W, KU on Río Indio, road to Las Minas, 593785 0988063 UTM, 115280; Mount Sapo, 7°5831N, 78°2143W, MCZ-­A USNM 565027–565028; Miraflores, 9°0N, 79°36W, 9179–9180; Ortiga Camp, 8°45N, 78°0W, FMNH USNM 53841; Nueva Gorgona, 8°33N, 79°52W, AMNH 170432, 170464; Río Abobi, Río Chucunaque, AMNH 69715–69719; Panama City (= Old Panama = Panama 40897; mouth of Río Canglon, Río Chucunaque, 8°20N, Viejo), 8°58N, 79°32W, AMNH 41068–41070, 52742, 77°46W, AMNH 40585–40587, UMMZ 123281(2), FMNH 60487, 170433, KU 116814–116815; Pipeline 124998–125002, 125003(2), 125004–125005, 132670; Road, 9°7N, 79°47W to 9°15N, 79°49W, UMMZ Río Chico, near Avelinos, 8°12N, 77°36W, AMNH 152934; Puerto La Chorrera, near, 8°52N, 79°43W; Río 39787, 40979–40984; near Río Membrillo, Río Chucu- Mamone (= Río Mamoní), El Capitán (= La Capitana), naque, 8°39N, 77°47W, KU 116816–116821; ~7 km 9°9N, 79°4W, FMNH 16657, USNM 53982; Río Silu- above Río Mortí, Río Chucunaque, 8°52N, 77°59W, KU ganti (= Río Chuluganti), 9°1313N, 78°489W, UMMZ 107223–107227, 107906; just below mouth of Río Mortí, 137745–137748; Río Tocumen, 9°1N, 79°23W, MCZ-­A Río Chucunaque, 8°50N, 77°59W, UMMZ 132663; in- 10140; Rousseau, 8°58N, 79°35W, KU 67959; Saboga Is- tersection of Río Chucunaque and Río Ucurganti, 8°25N, land, 10 m, 8°37N, 79°4W, MCZ-­A 2444, 99314–99319; 77°48W, USNM 140616–140619; Río Chucurti (= Sub- San Miguel (= Isla San Miguel), 8°22N, 78°55W, AMNH curtí), near camp, 8°45N, 77°57W, AMNH 40622, 6669, MCZ-­A 2424, 6901–6902, UMMZ 48076, USNM 40627; near mouth of Río Clarita, 8°6N, 77°28W, 58062; Summit Gardens, 9°4N, 79°39W, KU 107904– UMMZ 132665–132666; Río Jesusito, Sapo Mountains, 107905, UMMZ 167314; Tapia, 9°4N, 79°25W, AMNH 8°1N, 78°16W, MCZ-­A 9167; mouth of Río Membrillo, 18915–18917, 18920; Tocumen, 9°5N, 79°23W, USNM 8°39N, 77°47W, AMNH 40551–40554, Río Pita, 10 m, 147504; 8 km NE Tocumen, 9°6N, 79°20W, KU 107207– FMNH 67884, 67886–67892; camp above Río Sansan (= 107208; near Venado Beach, 8°54N, 79°36W, USNM Río Sansón), Río Chucunaque, 8°24N, 77°47W, AMNH 19336. San Blas: Camp Sasardí, 12 m, 9°0N, 77°48W, 40646–40647; Río Suchuti (= Río Sucutí = Río Subcurtí), KU 115275–115279. Veraguas: 3.4–5 km N Montijo, 8°45N, 77°57W, AMNH 40798; mouth of Río Tuquesa 7°59N, 81°4W, KU 107907–107908, 107198–107201. on Río Chucunaque, 8°23N, 77°47W, AMNH 40761– TRINIDAD. Nariva: Brickfield, 10°20N, 61°16W, 40765; Río Ucurganti, ~7 km above mouth, 30 m, 8°28N, FMNH 49661–49663. St. Patrick: Icacos Peninsula, 77°48W, KU 10725, 116822–116823; Santa Fe camp, USNM 306180; Icacos Point, 10°2N, 61°55W, USNM FMNH 170434–170435; Three Falls Creek (= Camp 287010; Laurier Pond near Siparia, 10°8N, 61°30W (co- Creek, Camp Townsend), 8°10N, 77°42W, AMNH ordinates for Siparia), FMNH 218782–218783. 44 • smithsonian contributions to zoology

VENEZUELA. Anzoategui: Carapá, near, 8°22N, Riecito, Acosta district, MCZ-A­ 26143; Río Socopito, 63°13W, USNM 80613–80614. Apuré: Hato La Gua- 80 km NW Carora, 470 m, 10°28N, 70°44W, USNM nota, 4 km W San Fernando de Apuré, 7°52N, 67°32W, 216731; entre Municipios Sucre y Petit, Sector La Ceiba, TCWC 45153, 45167. Aragua: Cuyagua, 10°29N, Río Mitare, a la altura del pueblo de El Paso, 11°6N, 67°42W, USNM 258124; El Limón, above Maracay, 69°55W, AMU 3234. Guárico: 8 km N, 2 km W Corozo 400 m, 10°18N, 67°38W, UMMZ 157138; Hacienda Pando, ~8°30N, 67°35W, TCWC 45852; Sosa, 9°20N, La Trinidad, near Maracay, 455 m, 10°17N, 67°37W, 67°12W, FMNH 125399, 176330. Mérida: Río Lim- FMNH 35988–35989; Lake Valencia, northern shore, ones, 8°51N, 71°27W, ULABG 5090, 5111–5113, 5170, 415 m, 10°11N, 67°45W, UMMZ 157102–157103; 5304. Miranda: Araguita, 1 km S, 10°13N, 66°28W, KU Maracay, 455 m, 10°15N, 67°36W, FMNH 190403, 166409–166410; Campo Grande de Caucagua, 10°17N, MCZ-­A 50703; south of highway between Maracay and 66°20W, MZUSP 58708–58711; Hacienda Bejuquero, Valencia, 13 km W Maracay, 420 m, 10°16N, 67°38W, 1 km south Río Chico, 1 m, 10°19N, 65°58W, USNM UMMZ 157104–157106; near Ocumare, 25 m, 10°28N, 216750–216753. Monagas: Caripito, Río Caripe, ~100 m, 67°46W, UMMZ 122371(2), 171686–171689; road 10°8N, 63°6W, AMNH 70650–70652; Escuela Granja between Ocumare (inland village) and Rancho Grande, Parcela, 8 km west-southwest­ Caripito, KU 117127. Por- 10°22N, 67°42W, UMMZ 157100; Parque Nacional tuguesa: Urbanización Andrés Eloy Blanco, Guanare, Henri Pittier (Estación Biologica Rancho Grande), 910– 180 m, 9°3N, 69°45W, ELM 1167. Sucre: Río Cocollar, 1,170 m, 10°22N, 67°41W, AMNH 70653, KU 132782, 10°10N, 63°47W, FMNH 17769–17771; Guaraúnos, MZUSP 8083, UMMZ 122372, 157101; near Turmero, ~100 m, 10°33N, 63°7W, KU 166404; Tocal, 11 km east of Maracay, 466 m, 10°14N, 67°29W (coordinates SSW Cumuna, 10°24N, 64°14W, KU 117122. Táchira: for Turmero), FMNH 27349. Bolívar: Ciudad Guayana, La Fria Pueblo Nuevo, 8°13N, 72°15W, UMMZ 55582. Parque Loefling, 100 m, 8°8N, 63°33W, USNM 229778; Trujillo: near El Divide, 28–30 km NW Valera, 90 m, Santo Tomé de Guayana (= San Felix de Guayana), 50 m, 9°31N, 70°44W, USNM 216754–216758. Yaracuy: El 8°2112N, 62°3910W, AMNH 81456–81457. Cara- Central, 10 km NW Urama, 25 m, 10°32N, 68°23W, bobo: El Central, 10 km NW, 25 m, 10°32N, 68°23W, USNM 216760; Palma Sola, 5 km SE, 10°36N, 68°34W, USNM 216759; Río San Esteban, 10°2841N, 68°115W, UMMZ 55581; 19 km NW of Urama, km 40, 25 m, UMMZ 55580. Distrito Federal: La Guaira, sea level, 10°37N, 68°24W, USNM 216727–216728, 216732– 10°36N, 66°56W, USNM 22539, 27793. Falcón: near 216749. Zulia: Encontrados, 10 m, 9°30N, 72°14W, La Pastora, 12 km ENE Mirimiri, 220 m, 11°12N, FMNH 2605; Setor Vera de Agua, km 12 on road El 68°38W, USNM 216729–216730; Municipio Mauroa, Vigía–Santa Barbara, 115 m, 8°422N, 71°437W, 22 km SW Guajiro (= Goajiro) by car, AMU 3290; Pauji, ULABG 5313. Acosta district, 11°1N, 68°34W, MCZ-A­ 25985–25988; Acknowledgments

e thank the following colleagues for assistance in examining specimens at their respective institutions (for institutional codes, see Sabaj Pérez, 2010) and/or responding to queries involving Wspecimens at their institutions: Andrés Acosta-­Galvis (Museo Universidad Javariana, Bogota), Ulisses Caramaschi (MNRJ), Ignacio De la Riva (MNCN), Marinus S. Hoogmoed (MPEG), Robert F. Inger (FMNH), John D. Lynch (ICNMHN), Charles W. Myers (AMNH), José P. Pombal, Jr. (MNRJ), Stephen P. Rogers (CM), Alan Resetar (FMNH), Gregory Schneider (UMMZ), Linda Trueb (KU), and Richard P. Vogt (Instituto Nacional de Pes- quisas da Amazônia, Manaus). The following colleagues generously provided help with difficult-to-­ ­pinpoint localities: Andrés Acosta-­Galvis (Museo Universidad Javariana, Bogota), Erik Ahlander (NHRM), James P. Bogart (University of Guelph), José Enrique Cas- tillo (IND), Maureen A. Donnelly (Florida International University), David Kizirian (AMNH), Enrique La Marca (Universidad de Los Andes, Mérida), John D. Lynch (ICNMHN), Charles W. Myers (AMNH), A. Stanley Rand (de- ceased, Smithsonian Tropical Research Institute), Alan Resetar (FMNH), José P. Rosado (MCZ), Jay M. Savage (Rana Dorada Enterprises, S.A.), John E. Sim- mons (KU), Neal Smith (Smithsonian Tropical Research Institute), and Paulo E. Vanzolini (MZUSP). Materials or material queries were generously made available or responded to for this project by César Barrio-Amarós­ (Fundación AndígenA, Mérida), Ig- nacio De la Riva (MNCN), William E. Duellman (KU), Roberto Ibáñez (Smith- sonian Tropical Research Institute), Travis La Duc (TNHC), Enrique La Marca (Universidad de Los Andes, Mérida), and John D. Lynch (Universidad Nacional de Colombia, Bogota). Lee-­Ann C. Hayek (National Museum of Natural History, Smithsonian In- stitution [NMNH-­SI]) provided statistical advice. Addison Wynn (NMNH-­SI) reworked Figure 9 for publication. The research was supported by National Science Foundation award DEB-­0342918.

Appendix 1: Observer Error Assessment for Morphological Data

Upper Shank, Outer Tarsus, and Sole of Foot Textures

After recording data on 1,002 specimens, it seemed that the amount of vari- ation within population samples was about the same as that among samples for the surface textures of the dorsal surface of the shank, outer surface of the tarsus, and sole of foot. An analysis was undertaken to determine whether there was meaningful intersample variation in the characters involved. Data were summarized for eight samples throughout the geographic distri- bution of the species complex: (1) 12 juveniles, 20 females, 4 males from Boca do Acre, Amazonas, Brazil; (2) 28 juveniles, 7 females, 1 male from Ilha do Maracá, Roraima, Brazil; (3) 10 juveniles, 4 females, 5 males from Hacienda El Ceibal, Bolívar, Colombia; (4) 30 juveniles, 3 females, 2 males from Beltrán, Cundina- marca, Colombia; (5) 23 juveniles, 4 females, 5 males from between Santa Marta and Mamatoco, Magdalena, Colombia; (6) 40 juveniles, 5 females, 3 males from a few localities along the Río Chucunaque, Darién, Panama; (7) 46 juveniles, 1 female, 1 male from Pakitza and Tambopata, Madre de Dios, Peru; (8) 10 juve- niles, 5 females, 3 males from 19 km northwest Urama, Falcón, Venezuela. Data were summarized for juveniles, females, and males for the following states of a surface shagreen: none, weak shagreen, and shagreen. These states were scored separately for the shank, tarsus, and foot. The shagreen texture is sandpaper-like,­ discernable only under magnification, with the skin surface just moist so the light reflects off the small protuberances comprising the shagreen. Shagreened surfaces seem to be sloughed off in specimens that are not well preserved. Data were also summarized for juveniles, females, and males for the following tubercle states: none, few, scattered, and many. These states were summarized separately for the shank, tarsus, and foot. The tubercles involved in well-­preserved specimens are small, black keratinized protuberances. In less than well-preserved­ specimens, the outer keratinized covering may be lost leaving (smaller) white tubercles. These structures are consistently interpretable only under magnification. The character states form a continuum (e.g., for shagreen, three states were coded as none, weak shagreen, and shagreen). If a given specimen was scored as 48 • smithsonian contributions to zoology

“none” in the first data-­taking session and “none” in the a 2-­step difference (all involved No Shagreen and Shagreen second data-­taking session, there is no difference in the states). For the sole of the foot surface, 92% were scored scoring between the two sessions for shagreen for the spec- as having the same state (No Shagreen) and 8% as having imen involved. If a given specimen was scored as “none” a 1-­step difference (all involved No Shagreen and Weak in the first data-­taking session and “weak shagreen” in the Shagreen states). These results indicate that observer error second data-taking­ session, that is considered to be a one-­ masks any differences that occur among individuals for step difference; if the specimen was scored as “none” in the states involved. The only conclusion that can be sup- the first data-­taking session and “shagreen” in the second ported with the large observer error is that the sole of the session, that is considered to be a two-­step difference. The foot generally lacks a shagreen in members of this species two-­step difference represents a larger observer error than complex and that most individuals have shagreened dorsal a one-step­ difference. The term step does not have any cla- shank and outer tarsal surfaces. distic connotation as used in the following analyses. Examination of the distribution of occurrences of Tubercles character states among the eight samples indicates that there are no occurrence differences among juveniles, fe- For the upper shank surface, 49% of the individuals males, and males within each locality. The distributions were scored as having the same state, 47% were scored of states are so obvious that statistical analyses are not as having a one-step­ difference (all involving the scattered required (data available from WRH upon request). The tubercles and many tubercles states), and 4% as having a data appeared as though there might be meaningful varia- two-­step difference (all involving the no tubercles and scat- tion among the localities for both the shagreen and tu- tered tubercles states). For the outer tarsal surface, 54% bercle characters for the three different surfaces involved. of the individuals were scored as having the same state, As the character states are defined qualitatively, subjective 39% with a one-­step difference (mostly involving the few interpretation could well vary from data-taking­ session to tubercles and scattered tubercles states), and 6% of the data-­taking session by the same observer. As there were individuals with two-step­ differences (all involving the no large time intervals between bouts of data taking for this tubercles and scattered tubercles states). For the sole of the project, the interpretations could well be quite different. foot surface, 37% of the individuals were scored as having To evaluate the impact of observer error, USNM specimens the same state, 60% were scored as having a one-step­ dif- for two of the samples that were involved in the first bout ference (mostly involving the no tubercles and few tuber- of data taking (Amazonas, Brazil, and Cundinamarca, Co- cles states), and 2% with a two-step­ difference (involving lombia) were rescored for the characters involved. Thus the no tubercles and scattered tubercles states). Observer each specimen had two character states recorded. For ex- error for the tubercle states has less of an impact than for ample, the upper shank surface on the first observation the shagreen states but is still considerable. Ninety-­four to might have been scored as “none” and when the second 98% of all error falls within a one-­step difference of the observation was made on the same specimen, it could have four-­state continuum of categories. All of the differences scored as “shagreened.” Alternatively, the first and second observed in distribution of states among the eight geo- evaluations of the same specimen could have been “none” graphic samples occurred between adjacent states, how- and “none.” The results are reported separately for the ever, indicating that the variation is not meaningful when two characters. observer error is accounted for. Owing to observer error and limited variation ob- Shagreen served in the eight geographic samples, it is clear that geo- graphic variation cannot be demonstrated convincingly For the upper shank surface, 33% of the individuals for the texture characters involved. Consequently, data for were scored as having the same state, 10% were scored as these characters were not taken after February 2005. having a 1-step­ difference (all of the differences involved The following characterization can be made for the were those between the Weak Shagreen and Shagreen Leptodactylus bolivianus complex. Shagreened surfaces states), and 57% of the individuals were scored as having are most commonly observed on the entire dorsal shank a 2-­step difference (all involved No Shagreen and Shagreen and outer tarsal surfaces. The sole of the foot usually does states). For the outer tarsal surface, 15% were scored as not have any shagreen; however, in the few cases where having the same state, 33% with a 1-step­ difference (all in- shagreen is present, it is limited to the proximal half of the volved No Shagreen and Weak Shagreen states), and 52% fifth metatarsal region of the sole of the foot. Almost all number 635 • 49 individuals have small black keratinized (or white) tuber- the basis of dissections made during the second session. cles rather evenly distributed on the dorsal surface of the These differences are not considered to be observer errors. shank and outer surface of the tarsus. The sole of the foot Specimen USNM 145715 (76 mm SVL) was scored as may lack tubercles entirely, although most specimens have an adult female in sessions 1 and 2 but a juvenile female them. When black (or white) tubercles are present, they in session 3. Reexamination of the specimen indicates that are typically smaller than those found on the dorsal shank with the dissection view available during sessions 1, 2, and and outer tarsal surfaces and characteristically occur only 3, only the ovary was visible. Further dissection was re- on the outer half of the sole of the foot. quired to expose an oviduct, which is curly; thus the speci- men is an adult female following the criteria used in this study. Methodology for Remaining Specimen USNM 145716 (71 mm SVL) was scored as Characters Exclusive of Male a juvenile female in sessions 1 and 3 and an adult female Secondary Sexual Characters in session 2. Reexamination of the specimen indicates that is has a straight oviduct and the ovaries are expanded con- The same series of 36 specimens from Beltrán, Cun- taining small white ova. The straight oviduct condition is dinamarca, Colombia evaluated for observer error in tex- used to define the juvenile female condition in this study. ture characters (preceding section) were used to evaluate Specimen USNM 145721 (63 mm SVL) was scored as observer error for sex, belly pattern, dorsal pattern, lip juvenile in session 1, a juvenile female in session 2, and a stripe pattern, posterior thigh pattern, upper shank pat- juvenile male in session 3. A fourth examination to deter- tern, longitudinal body folds, and male secondary sexual mine which of the results in sessions 1, 2, or 3 was correct characteristics. Data were originally recorded for these clearly indicated that the specimen has a straight oviduct specimens at the beginning of the study prior to 27 August with expanded ovaries with small white ova upon reexam- 2004. Character states were defined as written statements ination. The specimen is unquestionably a juvenile female; (e.g., for upper shank textures) or, where written state- there had to be a transcription recording error in session ments were not as clear as sketches, drawings were ren- 3, where the condition observed was not the condition re- dered. As data taking proceeded, if a character was similar corded during session 3. but not identical to a sketch, a brief statement character- Specimen USNM 145724 (59 mm SVL) was recorded izing the difference was added (Figure A1.1). Some pat- as a juvenile in session 1, a juvenile female in session 2, tern states were added after 27 August 2004. In theory, as and an adult female in session 3. Reexamination of the any new pattern was encountered during the study, a new specimen indicates that the oviduct is curly and the ova- standard was prepared for it. Thus any specimen that was ries are expanded with moderate sized white ova, an adult examined early in the study should not be scored as hav- female according to the criteria in this study. ing a standard subsequently added if the specimen were In conclusion, there are two kinds of observer error to be reexamined. In order to test this aspect along with detected in the reexamination process for determination observer error in general, the Beltran series was rescored of sex. The first is making different conclusions about on 29–30 August 2005. In order to determine how repeat- whether near-­adult sized females are adults or juveniles able the character states were scored over a short period of (5% scoring error). The second kind of error is a data time, the Beltrán series was scored a third time on 13–14 transcription error (1% error). September 2005. Belly Pattern Sex A total of 15 belly pattern standards were recorded for The scoring for sex was anticipated to be invariant the three Beltrán sample data-­taking sessions. Two pattern among the observation recording sessions with the excep- standards, added after the initial session, were scored with tion of a few of the larger juveniles that were not examined a 24% frequency in sessions 2 and 3. This indicates that internally the first session and scored simply as juveniles not all variation thought to be important was recognized but were examined internally the second session and were at the beginning of the study. recorded as either juvenile females or juvenile males. Only 14% of the standards were scored as the same Six specimens were scored differently between the first between sessions 1 and 2, and 17% between session 1 and and second and first and third data-recording­ sessions on 3; whereas 39% of the standards were identically scored 50 • smithsonian contributions to zoology

FIGURE A1.1. Example of design pattern. First page of belly pattern standards used in study to score character states. The original sheet had written annotations defining additional standards in relation to the pattern involved: A-­1, dark and light areas about equal; A-­2, mostly dark with few light spots, more light spots posteriorly; A-3­ as A-1,­ except darker to lighter gradient anteriorly-posteriorly;­ B-­1, mottling on posterior, almost uniform dark anteriorly; B-­2, uniformly mottled as on posterior; B-3,­ gradient of anterior pattern posteriorly; B-4,­ posterior pattern with anterior-­posterior gradient; B-­5, anterior pattern on entire belly; C-1,­ anterior mottling continuous across middle; C-2,­ almost lacking pattern; C-­3, uniformly darkly mottled (fine); C-­4, uniformly lightly mottled (fine); D-­1, mostly light; D-­2, indistinct anteriorly; E-­1 (only anterior por- tion filled in), fewer spots; E-­2, very few spots; F, virtually no pattern; F-­1, just a few small blotches lateral-­most; F-­2, uniformly dark; G-­1, very little in middle; G-­2, groups of melanophores instead of discrete marks; G-­3, pattern anteriorly only; H-­1, finer pattern; H-­2, almost uniform dark anteriorly; H-­3, anterior-­posterior gradient, lighter posteriorly; I-­1, encroachment only a bit laterally behind arms; I-2,­ pattern extends to posterior belly; J-­1, + a few [blotches] on lateral belly; J-­2, everywhere except center of belly; J-­3, light speckling over entire belly. number 635 • 51 between sessions 2 and 3. Thus there is more consistency mottled anteriorly only or with an anterior–posterior gra- in scoring over a shorter time period (sessions 2 and 3) dient, (2) belly boldly mottled anteriorly only or with an than for longer time periods between data taking (sessions anterior–posterior gradient over the entire belly, (3) belly 1–2, 1–3). speckled or mottled with the same intensity over the en- Given the low repeatability of scoring the 15 standards tire belly, (4) belly boldly mottled with the same intensity recorded for the Beltrán sample, the following two-­step over the entire belly, and (5) belly without a pattern (no procedure was used to combine the standards to maximize melanophores). These states are assumed to have around repeatability on scoring of the data. 50% repeatability over the entire study and around 75% The first step was to visually inspect the 15 standards repeatability within data-­taking sessions at single institu- and combine states that were most similar. This resulted in tions or for multiple institutions visited consecutively dur- recognition of seven states: (1) anterior mottling only, (2) ing the same trip. Thus any belly pattern differences in belly mottled with an anterior–posterior gradient, (3) uni- data comparisons must exceed 50% (for adequate sample form mottling over the entire belly, (4) belly with anterior sizes) to be meaningful. speckling only, (5) belly speckled with an anterior–pos- terior gradient, (6) belly uniformly speckled, and (7) no Dorsal Pattern pattern (no melanophores on belly). The Beltrán sample data were rescored using these seven states. This resulted Eight dorsal pattern standards were recorded during in 17% exact matches between session 1 and session 2, the three Beltrán sample data-taking­ sessions. The same 25% between session 1 and session 3, and 56% between standards were used for all sessions. One transcription session 2 and session 3. error was made where a nonexistent dorsal pattern stan- The second step was to examine the raw occurrence dard was recorded. data from step 1 to determine which further combinations Fifty-six­ percent of the standards were recorded as of character states would result in the greatest repeat- identical between sessions 1 and 2, 58% between ses- ability in the data set. Combining mottling and speckling sions 1 and 3, and 75% between sessions 2 and 3. There states did not appreciably improve repeatability. However, is greater consistency in scoring over a shorter time period combining these states with either of two other combina- (sessions 2–3) than over a longer time period (sessions tions did result in noticeable improvement. When mottling 1–2, 1–3). + speckling were combined with anterior only + anterior– The dorsal pattern standards were examined visu- posterior gradient states, the states were identical in 50% ally, and the most similar patterns were combined to see of the time between sessions 1 and 2 and between sessions if the increase in repeatability offset the loss of potential 1 and 3, and 80% between session 2 and 3 scorings. When information on variation. The following three states were mottling + speckling were combined with anterior–poste- recognized and used to recode the three session data: (1) rior gradient + uniform states, the states were identical dorsum uniform between interocular bar (when present) 42% of the time between session 1 and 2 data, 53 % be- and sacrum; (2) single chevron between interocular bar tween session 1 and 3 data, and 69% between session 2 and sacral chevron (when present) without additional and 3 data. spots lateral to dorsal chevrons; and (3) complex pattern Thus, for the Beltrán data, the following final states with two dorsal chevrons and spotting all over dorsum, are recognized: (1) belly speckled or mottled anteriorly chevrons often elongate and fused with each other and in- only or with an anterior–posterior gradient, (2) belly terorbital bar. The recoded data yield 77% exact match of speckled or mottled with the same intensity over the entire states between data for sessions 1 and 2, 67% between ses- belly, and (3) belly patternless (no melanophores). sion 1 and 3 data, and 82% between session 2 and 3 data. Finally, the total data set was evaluated visually using The increase in repeatability of the combined stan- the above results as a guideline. Thirty-­four belly pattern dards is considered worth the cost of losing information states occur in the total data set. Among the states that on variability of the dorsal patterns. However, further were not recorded in the Beltrán samples are two distinct reduction of the states would be too great a loss of in- features: spotted bellies and boldly mottled bellies. How- formation on variability. For dorsal pattern variation, any ever, there are very few times the spotted belly state was adequate size sample comparisons have to differ by more recorded. The variation found, including these new fea- than 33% in order to be meaningful. tures, result in the recognition of the following states used The three states defined above encompass the varia- in analyses of belly patterns: (1) belly speckled, spotted, or tion recorded in the total data set. 52 • smithsonian contributions to zoology

Lip Stripe either the session 2 or 3 data. Two transcription errors were made where a nonexistent standard was recorded. Two features were evaluated for the lip stripe: the Identical pattern standards were recorded 28% of the overall lip stripe pattern and whether a distinct light patch time between sessions 1 and 2, 31% between sessions 2 was present immediately below the eye. and 3, and 31% between sessions 2 and 3. The scoring Eighteen pattern states were recorded for the overall was quite consistent, but with high observer error in scor- lip stripe pattern in the three Beltrán sample data sets. ing, among sessions. Nine percent of the first and second session data elements The standards were examined visually to combine the were scored identically, 29% were identical between ses- most similar patterns and to minimize scoring differences sion 1 and 3 data, and 42% were identical between ses- found in the raw scoring results. The following five states sion 2 and 3 data. One standard was added after the first were recognized and used to recode the three session data: session, which was scored at a 3% occurrence in the com- (1) posterior thigh with a uniform or variable dispersion bined session 2 and 3 data. of melanophores (weakly to moderately mottled) over the The lip pattern standards were examined visually and entire surface of the posterior thigh; (2) posterior thigh combined to minimize mismatches in the original pattern with a darker patch with small, irregular light spots on scoring results. Four states were used to recode the data the distal half of the thigh; (3) posterior thigh with a dark for the three sessions: (1) a noticeable broad light stripe vermiculated pattern on the distal half of the thigh; (4) extending from the tip of the snout to through the com- posterior thigh pattern ranges from boldly mottled with missural gland, bordered below by continuous or broken large light irregular spots or vermiculations on a dark darker markings on the upper lip and either bordered background to the light markings being more extensive above by a dark line under the eye or not; (2) a narrow than the dark areas, the light markings sometimes co- light stripe from at least the nostril to the lower posterior alesced; and (5) posterior thigh uniformly dark with only end of the eye; (3) a narrow light stripe from below the a very few light irregular markings. Identical states of the front of the eye through the commissural gland; and (4) rescored data between session 1 and session 2 occurred central lip region uniform (about same color intensity as 69% of the time, 92% of the time between sessions 1 and background dorsal color) with or without a dark stripe 3, and 64% of the time between sessions 2 and 3. under the eye and/or dark bars/mottling on the upper lip. Patterns 2 and 3 defined above were only scored twice Identical pattern states were scored 53% of the time for each in the entire data set. Pattern 5 defined above (created the session 1 and 2 data, 69% for session 1 and session 3 from examination of specimens of the L. latrans complex) data, and 57% for session 2 and session 3 data. was scored once in the rescoring of the Beltrán sample Further reduction of states would result in very little specimens. In addition, two other patterns were scored variation in the data set. For the overall lip pattern varia- once in the entire data set. All of these single-­ or double-­ tion, any adequate sample size comparisons have to differ occurrence states are combined with their closest patterns by more than 50% in order to be meaningful. with the result that only two states are recognized for this Three conditions of expression of a light patch under character: (1) posterior thigh with a uniform or variable the eye (subocular light spot) were recorded in the three dispersion of melanophores (weakly to moderately mot- Beltrán data sets: (1) distinct, (2) indistinct, and (3) ab- tled) usually over entire surface of posterior thigh, and (2) sent. Ninety-four­ percent of the first and second session posterior thigh boldly mottled with large light irregular data states were scored identically, 94% were identical be- spots/vermiculations on a dark background to the light tween sessions 1 and 3 data, and 92% were identical be- marks being more extensive than the dark areas, the light tween sessions 2 and 3 data. The level of observer error is markings sometimes coalesced. considered acceptable. There are no further states for this Adequate sample size comparisons have to differ by character in the entire data set. For light subocular patch more than 33% in order to be meaningful. variation, adequate sample size comparisons have to differ by more than 10% to be meaningful. Dorsal Shank Pattern

Posterior Thigh Pattern Six patterns were scored in the three data-­recording sessions of the Beltrán sample. No patterns were added Thirteen posterior thigh pattern standards were scored after the first session. for the three Beltrán data-­recording sessions. One pattern Identical pattern standard states were recorded 42% was added after the first session but was not recorded in of the time in sessions 1 and 2 data, 25% in sessions 1 and number 635 • 53

3 data, and 69% in sessions 2 and 3 data. Scoring between Identical states were scored 39% of the time between sessions was quite inconsistent. session 1 and session 2 data, 44% between sessions 1 and Only minimal combining of pattern states was pos- 3, and 53% between sessions 2 and 3. sible by visual examination of the standards, resulting in The initial states were combined into two states: (1) recognition of the following three character states: (1) uni- lateral fold distinct or slightly interrupted from supratym- form, (2) short dark transverse bars extending less than panic fold to leg and (2) lateral fold distinct only in groin halfway across shank, and (3) at least one dark transverse region, indistinct overall, or indiscernible (probably due bar extending halfway or more across the shank. Identical to poor preservation). The rescored data between sessions states in the rescored data occurred 50% of the time in 1 and 2 were identical in 61% of the comparisons, 71% sessions 1 and 2 data, 31% in sessions 1 and 3 data, and between sessions 1 and 3, and 88% between sessions 2 72% in sessions 2 and 3 data. These values are only mar- and 3. The increase in repeatability is considered worth ginally better with respect to the values based on uncom- the loss of variation information. bined pattern character states. Further combining of states There is no further variation in this character in the would result in little variation of the upper shank patterns. entire data set. Differences in lateral fold states must ex- There is no additional variation for this character in the ceed 40% among adequate size samples to be meaningful. entire data set. Adequate sample size comparisons have to The new states are used in subsequent analyses. differ by more than 70% in order to be meaningful. There are two extremes of repeatability of character states represented by belly and lip patterns at one extreme Dorsolateral Folds and the light subocular patch at the other. The belly and lip patterns involve several features that together repre- Six states were recognized in the Beltrán specimen sent a total gestalt. In retrospect, WRH focused on dif- data. One state was added after the first data-­recording ferent components of the belly and lip pattern standards session, which was recorded in 11% of the session 2 and at different times. For example, during one session more 3% of the session 3 data. attention might have been focused on the nature of the Fifty percent of the specimens were recorded with markings on the upper lip adjacent to the jaw; during an- identical states between sessions 1 and 2, 47% between other session the focus might have been on how distinc- sessions 1 and 3, and 53% between sessions 2 and 3. tive the broad light stripe was between the upper lip and The initial states were combined into two states to de- the eye; and yet another session may have focused more termine whether the increase in repeatability was worth the on the nature of the dark markings immediately below loss of information concerning variation: (1) dorsolateral the eye. Such unintentional shifts in focus could explain fold distinct and (a) continuous from supratympanic fold why there was more consistency in observations made to end of body, or (b) with a brief interruption of the fold, over short periods of time and much greater inconsistency or (c) the fold is distinct only from the supratympanic fold when data were taken after a long hiatus between ses- to the sacral region; and (2) dorsolateral fold with several sions. In contrast, the light subocular patch is a single interruptions to indiscernible (including lack of fold prob- feature for which the variation was recorded much more ably due to poor preservation). The rescored comparisons consistently. resulted in 78% exact matches between data sessions 1 There are at least two ways that complex patterns and 2, 65% between sessions 1 and 3, and 76% between could be analyzed differently than was done in this study sessions 2 and 3. The gain in repeatability is considered to minimize observer error. The first would be to take all worth the loss of information concerning variability. the data in one continuous session, where on a day-to-­ ­day The redefined states encompass all the variation ob- basis, there would likely be carryover of which features served in the entire data set. Differences in dorsolateral in complex patterns were being focused on. The second fold states must exceed 33% in adequate sample sizes to would be to pick a series of specimens that likely encom- be meaningful. pass the range of variation in the entire study sample and prepare a series of sketches that represent that range of Lateral Folds variation for each general character to be evaluated, such as belly pattern. On the basis of those sketches, individual Six states were recognized in the three data-taking­ ses- features could be identified and character states defined sions. One state was added after the first session and was that would then be used to evaluate the entire data set. recorded in 3% of the individuals for both the session 2 The second approach would seem more effective to mini- and 3 data. mize observer error. 54 • smithsonian contributions to zoology

Male Secondary patch of keratinized tubercles, and (3) male chest with a Sexual Characters central patch and two elongate lateral patches of keratin- ized tubercles just posterior to the central patch. There are only two adult males in the Beltrán sample. Male Thumb. (1) Male thumb with one The only variation among the three data sessions was a weakly chisel-shaped­ keratinized spine, (2) male thumb one-­step difference out of a four-state­ range in evaluation with one well-­developed chisel-­shaped keratinized spine, of the degree of male arm hypertrophy. and (3) male thumb with two round keratinized spines. The male secondary sexual characteristics can be The spines have keratinized sheaths at least during the defined and described in terms of individual features as reproductive season. Spines lacking sheaths still dem- follows. onstrate the differences between the chisel-shaped­ and Male Arm. (1) Upper arm not hypertrophied rounded features. and (2) upper arm hypertrophied. The above definitions would be expected to have the Male Lower Jaw. (1) Male lower jaw lack- same repeatability as those for the light subocular patch ing keratinized tubercles and (2) male lower jaw with ke- character. Erring on the conservative side, adequate sam- ratinized tubercles. ple size comparisons for the male secondary sexual char- Male Chest. (1) Male chest lacking a patch acters should differ by more than 15% to be meaningful. of keratinized tubercles, (2) male chest with a central Appendix 2. Molecular Sample Locality Data

LEPTODACTYLUS BOLIVIANUS COMPLEX LOCALITY DATA USED IN MOLECULAR ANALYSIS

Leptodactylus bolivianus. USNM 268966 (USFS 152368)—PERU: Madre de Dios, Tambopata Reserve, 12°50S, 69°17W, 66 m (collection of Jiri Moravec); BOLIVIA: Pando, settlement of Nacebe, 11°00S, 67°25W. Leptodactylus guianensis. AMNH145130–131 (JC 7056–57)—GUYANA: Corontyne Region, Dubulay Ranch on the Berbice River, 5°405N, 57°5132W, UC 128, 130; BRAZIL: Roraima, proximity of Vila Surumu, 04°1140.9N, 60°4725.1W. Leptodactylus insularum. MUJ 2187—COLOMBIA: no other data. JDL 24816—COLOMBIA: Sucre, San Marcos, vereda La Florida, 8°37N, 75°11W, 45 m. JDL 24887—COLOMBIA: San Marcos, finca Crocodilia, 8°36N, 75°09W, 34 m. JDL 26573, JDL 26591—COLOMBIA: Cordoba; Lorica, esta- cion pisicola CVS, 9°13N, 75°50W, 20 m. CH 4955–56—PANAMA: Panamá, Río Indio, camino hacia Las Minas. ULABG 5111–13, 5304—VENEZUELA: Mérida; Río Limones.

References

Barbour, T. 1906. Vertebrata from the Savanna of Panama. Reptilia and Amphibia. Bul- letin of the Museum of Comparative Zoology at Harvard College, 46:224–229. Boulenger, G. A. 1898. A List of the Reptiles and Batrachians Collected by the Late Prof. L. Balzan in Bolivia. Annali del Museo Civico di Storia Naturale di Genova, Series 2, 19:128–133. Capocaccia, L. 1957. Catalogo dei Tipi di Anfibi del Museo Civico di Storia Naturale di Genova. Annali del Museo Civico di Storia Naturale di Genova, 69:208–222. Charif, R. A., C. W. Clark, and K. M. Fristrup. 2004. Raven 1.2 User’s Manual. Ithaca, N. Y.: Cornell Laboratory of Ornithology. Duellman, W. E. 1997. Amphibians of La Escalera Region, Southeastern Venezuela: Tax- onomy, Ecology, and Biogeography. Scientific Papers, Natural History Museum, The University of Kansas, 2:1–52. ———. 2005. Cusco Amazónico. The Lives of Amphibians and Reptiles in an Amazo- nian Rainforest. Ithaca, N.Y.: Cornell University Press. Engelman, L., S. N. Badashah, and R. V. Nath. 2004. “Discriminant analysis.” In Statis- tics, ed. Anonymous, pp. I301–I358. Richmond, Va.: SYSTAT Software, Inc. Fabrezi, M., and P. Alberch. 1996. The Carpal Elements of Anurans. Herpetologica, 52:188–204. Felsenstein, J. 1985. Confidence Limits on Phylogenies: An Approach Using the Boot- strap. Evolution, 39:783–791. doi:10.2307/2408678. Fouquette, M. J., Jr. 1960. Call Structure in Frogs of the Family Leptodactylidae. The Texas Journal of Science, 12:201–215. Gosner, K. L. 1960. A Simplified Table for Staging Anuran Embryos and Larvae with Notes on Identification. Herpetologica, 16:183–190. Hayek, L.-A.­ C., W. R. Heyer, and C. Gascon. 2001. Frog Morphometrics: A Cautionary Tale. Alytes, 18:153–177. Heyer, W. R. 1969. Studies on Frogs of the Genus Leptodactylus (Amphibia, Leptodac- tylidae). V. Taxonomic Notes on L. latinasus, rhodonotus, romani, and wuchereri. Herpetologica, 25:1–8. ———. 1970 (“1968”). Studies on the Genus Leptodactylus (Amphibia: Leptodactyli- dae). II. Diagnosis and Distribution of the Leptodactylus of Costa Rica. Revista de Biología Tropical, 16:171–205. ———. 1994. Variation within the Leptodactylus podicipinus-­wagneri Complex of Frogs (Amphibia: Leptodactylidae). Smithsonian Contributions to Zoology, No. 546. Washington, D.C.: Smithsonian Institution Press. ———. 2005. Variation and Taxonomic Clarification of the Large Species of the Lep- todactylus pentadactylus Species Group (Amphibia: Leptodactylidae) from Middle 58 • smithsonian contributions to zoology

America, Northern South America, and Amazonia. Arqui- och Vitterhets—Samhälles Handlingar Sjätte Följden, Series vos de Zoologia, 37:269–348. B, 1(4):1–71. Heyer, W. R., A. S. Rand, C. A. G. Cruz, O. L. Peixoto, and Palumbi, S. R. 1996. Nucleic Acids II: The Polymerase Chain C. E. Nelson. 1990. Frogs of Boracéia. Arquivos de Zoolo- Reaction. In Molecular Systematics [2nd edition], ed. D. M. gia, 31:231–410. Hillis, C. Moritz, and B. K. Mable, pp. 205–247. Sunder- Ibáñez D., R., A. S. Rand, M. J. Ryan, and C. A. Jaramillo A. land, Mass.: Sinauer Associates. 1999. Vocalizaciones de ranas y sapos del Monumento Posada, D., and K. A. Crandall. 1998. Modeltest: Testing the Natural Barro Colorado, Parque Nacional Soberanía y Model of DNA Substitution. Bioinformatics, 14:817–818. áreas adyacentes. Vocalizations of frogs and toads from doi:10.1093/bioinformatics/14.9.817. Barro Colorado Nature Monument, Soberania National Sabaj Pérez, M. H., ed. 2010. Standard Symbolic Codes for Insti- Park and adjacent areas. CD-ROM.­ Panama City, Panama: tutional Resource Collections in Herpetology and Ichthyol- Fundación Natura, Circulo Herpetológico de Panamá, and ogy: An Online Reference. Version 2.0 (8 November 2010). Smithsonian Tropical Research Institute. American Society of Ichthyologists and Herpetologists, Lavilla, E. O., J. A. Langone, U. Caramaschi, W. R. Heyer, and R. Washington, D.C. Accessible at http://www.asih.org/. O. de Sá. 2010. The identification of Rana ocellata Linnaeus, Swofford, D. L. 2002. PAUP*. Phylogenetic Analysis Using Par- 1758. Nomenclatural impact on the species currently known simony (* and Other Methods). Version 4.0b10. Sunder- as Leptodactylus ocellatus (Leptodactylidae) and Osteopilus land, Mass.: Sinauer Associates. brunneus (Gosse, 1851) (Hylidae). Zootaxa, 2346:1–16. Swofford, D. L., G. J. Olsen, P. J. Waddell, and D. M. Hillis. Márquez, R., I. J. De la Riva, J. Bosch, and E. Matheu, eds. 2002. 1996. Phylogenetic Inference. In Molecular Systematics Guía Sonora de las Ranas y Sapos de Bolivia. Sounds of Frogs [2nd edition], ed. D. M. Hillis, C. Moritz, and B. K. Mable, and Toads of Bolivia. CD-ROM.­ Madrid: Alosa and Museo pp. 407–514. Sunderland, Mass.: Sinauer Associates. Nacional de Ciencias Naturales, Fonoteca Zoológica. Thompson, J, D., T. J. Gibson, F. Plewniak, F. Jeanmougin, McDiarmid, R. W., and R. Altig, eds. 1999. Tadpoles. The Biol- and D. G. Higgins. 1997. The ClustalX Windows Inter- ogy of Anuran Larvae. Chicago Ill.: University of Chicago face: Flexible Strategies for Multiple Sequence Alignment Press. Aided by Quality Analysis Tools. Nucleic Acids Research, Melin, D. 1941. Contributions to the Knowledge of the Am- 24:4876–4882. doi:10.1093/nar/25.24.4876. phibia of South America. Göteborgs Kungliga Vetenskaps— requirements for smithsonian series publication

All manuscripts are reviewed for adher- Art must not be embedded in the main text. ence to the SISP Manuscript Preparation and Style Guide for Authors (available on the “Submis- Figures must be original and submitted as individual TIFF sions” page at www.scholarlypress.si.edu). Manuscripts not or EPS files. Resolution for art files must be at least 300 dpi in compliance will be returned to the author. Manuscripts in- for grayscale and color images and at least 1200 dpi for line tended for publication in the Contributions Series are evalu- art. Electronic images should measure no more than 100% ated by a content review board and undergo substantive peer and no less than 75% of final size when published. JPG files review. Accepted manuscripts are submitted for funding ap- will not be accepted. Color images significantly increase proval and scheduling to the Publications Oversight Board. costs so should be included only if required. Funding for color art is subject to approval by SISP and the Publications Minimum manuscript length is thirty manuscript Oversight Board. pages. If a manuscript is longer than average, an appropriate length will be determined during peer review and evaluation taxonomic keys in natural history papers should use by the Content Review Board. Authors may be asked to edit the aligned-couplet form for zoology. If cross referencing is manuscripts that are determined to be too long. required between key and text, do not include page references within the key but number the keyed-out taxa, using the same Text must be prepared in a recent version of Microsoft numbers with their corresponding heads in the text. Word; use a Times font in 12 point for regular text; be double spaced; and have 1" margins. Each chapter/section must be synonomy in zoology must use the short form (tax- saved in a separate file. on, author, year:page), with full reference at the end of the paper under “References.” Required elements are title page, abstract page, table of contents, main text, and reference section. See the SISP In-text references should be used rather than biblio- Manuscript Preparation and Style Guide for Authors for the graphic notes and should follow the author-date system in the order of all elements. following format: “(author last name, year)” or “. . . author (year)”; “(author, year:page used within the text)” or “. . . Headings should be styled so different levels of headings are author (year:page).” A full citation should be included in a distinct from each other and so the organization of the manu- “References” section. script is clear. Insert one line space above and one line space below all headings. Endnotes are to be used in lieu of footnotes and should be keyed manually into a separate MS Word file, in a sec- Front matter should include title page, abstract page, tion titled “Notes”. Notes should not contain bibliographic and table of contents. All other sections are optional. Ab- information. Manually type superscript numerals in text and stracts must not exceed 300 words. Table of contents should use full-sized numerals at the beginning of each note in the include A-, B-, and C-level headings. “Notes” section. SISP will determine the best placement of the notes section, either at the end of each chapter or at the end Tables (numbered, with captions, stubs, rules) should be of the main text. submitted in separate MS Word files; should include foot- notes, if appropriate; should have rules only at top, bottom, References should be in alphabetical order, and in chron- and beneath column heads. Print outs of each table should ological order for same-author entries. Each reference should accompany the manuscript to ensure correct layout of data. be cited at least once in main text. Complete bibliographic Tabulations within running text should not be numbered or information must be included in all citations (e.g., author/edi- formatted like formal tables, and should be included in the tor, title, subtitle, edition, volume, issue, pages, figures). For text of the manuscript. books, place of publication and publisher are required. For journals, use the parentheses system for volume(number):pag- Figure captions should be provided in a separate MS ination [e.g., “10(2):5–9”]. Do not use “et al.”; all authors/ Word file. editors should be included in reference citations. In titles, capitalize first word, last word, first word after colon, and all Figures (e.g., photographs, line art, maps) should be num- other words except articles, conjunctions, and prepositions. bered sequentially (1, 2, 3, etc.) in the order called out; be Examples of the most common types of citations are provided placed throughout text, not at end of manuscript; have all in the SISP Manuscript Preparation and Author Style Guide. components of composites lettered with lowercase letters and described in the caption; include a scale bar or scale descrip- For questions regarding the guidelines, please email SISP at tion, if appropriate; include any legends in or on the figure [email protected]. rather than in a caption.